Our work aims to estimate the camera motion mounted on the head of a mobile robot or a moving object from RGB-D images in a static scene. The problem of motion estimation is transformed into a nonlinear least squares function. Methods for solving such problems are iterative. Various classic methods gave an iterative solution by linearizing this function. We can also use the metaheuristic optimization method to solve this problem and improve results. In this paper, a new algorithm is developed for visual odometry using a sequence of RGB-D images. This algorithm is based on a genetic algorithm. The proposed iterative genetic algorithm searches using particles to estimate the optimal motion and then compares it to the traditional methods. To evaluate our method, we use the root mean square error to compare it with the based energy method and another metaheuristic method. We prove the efficiency of our innovative algorithm on a large set of images.
Leader Follower Formation Control of Ground Vehicles Using Dynamic Pixel Coun...ijma
This paper deals with leader-follower formations of non-holonomic mobile robots, introducing a formation
control strategy based on pixel counts using a commercial grade electro optics camera. Localization of the
leader for motions along line of sight as well as the obliquely inclined directions are considered based on
pixel variation of the images by referencing to two arbitrarily designated positions in the image frames.
Based on an established relationship between the displacement of the camera movement along the viewing
direction and the difference in pixel counts between reference points in the images, the range and the angle
estimate between the follower camera and the leader is calculated. The Inverse Perspective Transform is
used to account for non linear relationship between the height of vehicle in a forward facing image and its
distance from the camera. The formulation is validated with experiments.
Stereo Correspondence Algorithms for Robotic Applications Under Ideal And Non...CSCJournals
The use of visual information in real time applications such as in robotic pick, navigation, obstacle avoidance etc. has been widely used in many sectors for enabling them to interact with its environment. Robotics require computationally simpler and easy to implement stereo vision algorithms that will provide reliable and accurate results under real time constraint. Stereo vision is a less expensive, passive sensing technique, for inferring the three dimensional position of objects from two or more simultaneous views of a scene and there is no interference with other sensing devices if multiple robots are present in the same environment. Stereo correspondence aims at finding matching points in the stereo image pair based on Lambertian criteria to obtain disparity. The correspondence algorithm will provide high resolution disparity maps of the scene by comparing two views of the scene under the study. By using the principle of triangulation and with the help of camera parameters, depth information can be extracted from this disparity .Since the focus is on real-time application, only the local stereo correspondence algorithms are considered. A comparative study based on error and computational costs are done between two area based algorithms. Evaluation of Sum of absolute Difference algorithm, which is less computationally expensive, suitable for ideal lightening condition and a more accurate adaptive binary support window algorithm that can handle of non-ideal lighting conditions are taken for this study. To simplify the correspondence search, rectified stereo image pairs are used as inputs.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...cscpconf
1) The document proposes a multiple region of interest (ROI) tracking algorithm for non-rigid objects using Demon's algorithm and a pyramidal approach.
2) A pyramidal implementation of Demon's algorithm improves computational efficiency and accuracy of tracking by calculating displacement fields at multiple image resolutions.
3) The algorithm is applied to track non-rigid ROIs in laparoscopy videos, which could help surgeons perform minimal invasive surgery.
Multiple region of interest tracking of non rigid objects using demon's algor...csandit
In this paper we propose an algorithm for tracking multiple ROI (region of interest) undergoing
non-rigid transformations. Demon's algorithm based on the idea of Maxwell's demon, has been
applied here to estimate the displacement field for tracking of multiple ROI. This algorithm
works on pixel intensities of the sequence of images thus making it suitable for tracking
objects/regions undergoing non-rigid transformations. We have incorporated a pyramid-based
approach for demon's algorithm computations of displacement field, which leads to significant
reduction in the convergence speed and improvement in the accuracy. This algorithm is applied
for tracking non-rigid objects in laproscopy videos which would aid surgeons in Minimal
Invasive Surgery (MIS).
An automatic algorithm for object recognition and detection based on asift ke...Kunal Kishor Nirala
This document presents an automatic algorithm for object recognition and detection based on ASIFT keypoints. The algorithm combines affine scale invariant feature transform (ASIFT) and a region merging algorithm. ASIFT is used to extract keypoints from a training image of the object. These keypoints are then used instead of user markers in a region merging algorithm to recognize and detect the object with full boundary in other images. Experimental results show the method is efficient and accurate at recognizing and detecting objects.
This document compares the accuracy of determining volumes using close range photogrammetry versus traditional methods. It presents a case study where the volume of a test field was calculated using both approaches. Using traditional methods with 425 control points, the volume was calculated as 221475.14 m3 using trapezoidal rules, 221424.52 m3 using Simpson's rule, and 221484.05 m3 using Simpson's 3/8 rule. Using close range photogrammetry with 42 control points and 574 generated points, the volume was calculated as 215310.60 m3 using trapezoidal rules, 215300.43 m3 using Simpson's rule, and 215304.12 m3 using Simpson's 3
APPLYING R-SPATIOGRAM IN OBJECT TRACKING FOR OCCLUSION HANDLINGsipij
Object tracking is one of the most important problems in computer vision. The aim of video tracking is to extract the trajectories of a target or object of interest, i.e. accurately locate a moving target in a video sequence and discriminate target from non-targets in the feature space of the sequence. So, feature descriptors can have significant effects on such discrimination. In this paper, we use the basic idea of many trackers which consists of three main components of the reference model, i.e., object modeling, object detection and localization, and model updating. However, there are major improvements in our system. Our forth component, occlusion handling, utilizes the r-spatiogram to detect the best target candidate. While spatiogram contains some moments upon the coordinates of the pixels, r-spatiogram computes region-based compactness on the distribution of the given feature in the image that captures richer features to represent the objects. The proposed research develops an efficient and robust way to keep tracking the object throughout video sequences in the presence of significant appearance variations and severe occlusions. The proposed method is evaluated on the Princeton RGBD tracking dataset considering sequences with different challenges and the obtained results demonstrate the effectiveness of the proposed method.
Leader Follower Formation Control of Ground Vehicles Using Dynamic Pixel Coun...ijma
This paper deals with leader-follower formations of non-holonomic mobile robots, introducing a formation
control strategy based on pixel counts using a commercial grade electro optics camera. Localization of the
leader for motions along line of sight as well as the obliquely inclined directions are considered based on
pixel variation of the images by referencing to two arbitrarily designated positions in the image frames.
Based on an established relationship between the displacement of the camera movement along the viewing
direction and the difference in pixel counts between reference points in the images, the range and the angle
estimate between the follower camera and the leader is calculated. The Inverse Perspective Transform is
used to account for non linear relationship between the height of vehicle in a forward facing image and its
distance from the camera. The formulation is validated with experiments.
Stereo Correspondence Algorithms for Robotic Applications Under Ideal And Non...CSCJournals
The use of visual information in real time applications such as in robotic pick, navigation, obstacle avoidance etc. has been widely used in many sectors for enabling them to interact with its environment. Robotics require computationally simpler and easy to implement stereo vision algorithms that will provide reliable and accurate results under real time constraint. Stereo vision is a less expensive, passive sensing technique, for inferring the three dimensional position of objects from two or more simultaneous views of a scene and there is no interference with other sensing devices if multiple robots are present in the same environment. Stereo correspondence aims at finding matching points in the stereo image pair based on Lambertian criteria to obtain disparity. The correspondence algorithm will provide high resolution disparity maps of the scene by comparing two views of the scene under the study. By using the principle of triangulation and with the help of camera parameters, depth information can be extracted from this disparity .Since the focus is on real-time application, only the local stereo correspondence algorithms are considered. A comparative study based on error and computational costs are done between two area based algorithms. Evaluation of Sum of absolute Difference algorithm, which is less computationally expensive, suitable for ideal lightening condition and a more accurate adaptive binary support window algorithm that can handle of non-ideal lighting conditions are taken for this study. To simplify the correspondence search, rectified stereo image pairs are used as inputs.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...cscpconf
1) The document proposes a multiple region of interest (ROI) tracking algorithm for non-rigid objects using Demon's algorithm and a pyramidal approach.
2) A pyramidal implementation of Demon's algorithm improves computational efficiency and accuracy of tracking by calculating displacement fields at multiple image resolutions.
3) The algorithm is applied to track non-rigid ROIs in laparoscopy videos, which could help surgeons perform minimal invasive surgery.
Multiple region of interest tracking of non rigid objects using demon's algor...csandit
In this paper we propose an algorithm for tracking multiple ROI (region of interest) undergoing
non-rigid transformations. Demon's algorithm based on the idea of Maxwell's demon, has been
applied here to estimate the displacement field for tracking of multiple ROI. This algorithm
works on pixel intensities of the sequence of images thus making it suitable for tracking
objects/regions undergoing non-rigid transformations. We have incorporated a pyramid-based
approach for demon's algorithm computations of displacement field, which leads to significant
reduction in the convergence speed and improvement in the accuracy. This algorithm is applied
for tracking non-rigid objects in laproscopy videos which would aid surgeons in Minimal
Invasive Surgery (MIS).
An automatic algorithm for object recognition and detection based on asift ke...Kunal Kishor Nirala
This document presents an automatic algorithm for object recognition and detection based on ASIFT keypoints. The algorithm combines affine scale invariant feature transform (ASIFT) and a region merging algorithm. ASIFT is used to extract keypoints from a training image of the object. These keypoints are then used instead of user markers in a region merging algorithm to recognize and detect the object with full boundary in other images. Experimental results show the method is efficient and accurate at recognizing and detecting objects.
This document compares the accuracy of determining volumes using close range photogrammetry versus traditional methods. It presents a case study where the volume of a test field was calculated using both approaches. Using traditional methods with 425 control points, the volume was calculated as 221475.14 m3 using trapezoidal rules, 221424.52 m3 using Simpson's rule, and 221484.05 m3 using Simpson's 3/8 rule. Using close range photogrammetry with 42 control points and 574 generated points, the volume was calculated as 215310.60 m3 using trapezoidal rules, 215300.43 m3 using Simpson's rule, and 215304.12 m3 using Simpson's 3
APPLYING R-SPATIOGRAM IN OBJECT TRACKING FOR OCCLUSION HANDLINGsipij
Object tracking is one of the most important problems in computer vision. The aim of video tracking is to extract the trajectories of a target or object of interest, i.e. accurately locate a moving target in a video sequence and discriminate target from non-targets in the feature space of the sequence. So, feature descriptors can have significant effects on such discrimination. In this paper, we use the basic idea of many trackers which consists of three main components of the reference model, i.e., object modeling, object detection and localization, and model updating. However, there are major improvements in our system. Our forth component, occlusion handling, utilizes the r-spatiogram to detect the best target candidate. While spatiogram contains some moments upon the coordinates of the pixels, r-spatiogram computes region-based compactness on the distribution of the given feature in the image that captures richer features to represent the objects. The proposed research develops an efficient and robust way to keep tracking the object throughout video sequences in the presence of significant appearance variations and severe occlusions. The proposed method is evaluated on the Princeton RGBD tracking dataset considering sequences with different challenges and the obtained results demonstrate the effectiveness of the proposed method.
Data-Driven Motion Estimation With Spatial AdaptationCSCJournals
The pel-recursive computation of 2-D optical flow raises a wealth of issues, such as the treatment of outliers, motion discontinuities and occlusion. Our proposed approach deals with these issues within a common framework. It relies on the use of a data-driven technique called Generalised Cross Validation to estimate the best regularisation scheme for a given pixel. In our model, the regularisation parameter is a general matrix whose entries can account for different sources of error. The motion vector estimation takes into consideration local image properties following a spatially adaptive approach where each moving pixel is supposed to have its own regularisation matrix. Preliminary experiments indicate that this approach provides robust estimates of the optical flow.
Stereo Vision Distance Estimation Employing Canny Edge Detector with Interpol...ZaidHussein6
This document summarizes a research paper that proposes a stereo vision algorithm called the Canny Block Matching Algorithm (CBMA) to estimate distance from stereo images. CBMA uses the Canny edge detector to extract edges from images and block matching with Sum of Absolute Difference (SAD) to determine disparity maps and reduce processing time. The algorithm was tested on stereo image pairs and achieved an error reduction of about 2% and processing time reduction compared to other methods. Interpolation techniques including bilinear, 1st order polynomial and 2nd order polynomial were also evaluated to enhance the output images and further reduce errors.
Goal location prediction based on deep learning using RGB-D camerajournalBEEI
In the navigation system, the desired destination position plays an essential role since the path planning algorithms takes a current location and goal location as inputs as well as the map of the surrounding environment. The generated path from path planning algorithm is used to guide a user to his final destination. This paper presents a proposed algorithm based on RGB-D camera to predict the goal coordinates in 2D occupancy grid map for visually impaired people navigation system. In recent years, deep learning methods have been used in many object detection tasks. So, the object detection method based on convolution neural network method is adopted in the proposed algorithm. The measuring distance between the current position of a sensor and the detected object depends on the depth data that is acquired from RGB-D camera. Both of the object detected coordinates and depth data has been integrated to get an accurate goal location in a 2D map. This proposed algorithm has been tested on various real-time scenarios. The experiments results indicate to the effectiveness of the proposed algorithm.
This project is to retrieve the similar geographic images from the dataset based on the features extracted.
Retrieval is the process of collecting the relevant images from the dataset which contains more number of
images. Initially the preprocessing step is performed in order to remove noise occurred in input image with
the help of Gaussian filter. As the second step, Gray Level Co-occurrence Matrix (GLCM), Scale Invariant
Feature Transform (SIFT), and Moment Invariant Feature algorithms are implemented to extract the
features from the images. After this process, the relevant geographic images are retrieved from the dataset
by using Euclidean distance. In this, the dataset consists of totally 40 images. From that the images which
are all related to the input image are retrieved by using Euclidean distance. The approach of SIFT is to
perform reliable recognition, it is important that the feature extracted from the training image be
detectable even under changes in image scale, noise and illumination. The GLCM calculates how often a
pixel with gray level value occurs. While the focus is on image retrieval, our project is effectively used in
the applications such as detection and classification.
This document proposes a method for change detection in images that combines Change Vector Analysis, K-Means clustering, Otsu thresholding, and mathematical morphology. It involves detecting intensity changes using CVA, segmenting the difference image using K-Means, calculating a threshold with Otsu's method, applying the threshold and morphological operations, and comparing results to other change detection techniques. Experimental results on medical and other images show the proposed method achieves satisfactory change detection with fewer errors compared to other methods.
3D Reconstruction from Multiple uncalibrated 2D Images of an ObjectAnkur Tyagi
3D reconstruction is the process of capturing the shape and appearance of real objects. In this project we are using passive methods which only use sensors to measure the radiance reflected or emitted by the objects surface to infer its 3D structure.
This document proposes using dual back-to-back Kinect sensors mounted on a robot to capture a 3D model of a large indoor scene. Traditionally, one Kinect is slid across an area, but this requires prominent features and careful handling. The dual Kinect setup requires calibrating the relative pose between the sensors. Since they do not share a view, traditional calibration is not possible. The authors place a dual-face checkerboard on top with a mirror to enable each Kinect to view the same calibration object. This allows estimating the pose between the sensors using a mirror-based algorithm. After capturing local models, the two Kinect views can be merged into a combined 3D model with a larger field of view.
This document describes a laser distance measurement system using a webcam. It consists of a laser transmitter and webcam receiver. The laser pulse is reflected off an object and received by the webcam. Software calculates the distance based on the time of flight. The system achieves high accuracy of ±3cm. It calibrates the system using test measurements to determine the relationship between pixel location of the laser dot and actual distance. This allows accurate distance measurements within a few percent of error out to over 2 meters. Potential improvements discussed are using a laser line instead of dot for more data points and a green laser for better visibility.
Gabor filter is a powerful way to enhance biometric images like fingerprint images in order to extract correct features from these images, Gabor filter used in extracting features directly asin iris images, and sometimes Gabor filter has been used for texture analysis. In fingerprint images The even symmetric Gabor filter is contextual filter or multi-resolution filter will be used to enhance fingerprint imageby filling small gaps (low-pass effect) in the direction of the ridge (black regions) and to increase the discrimination between ridge and valley (black and white regions) in the direction, orthogonal to the ridge, the proposed method in applying Gabor filter on fingerprint images depending on translated fingerprint image into binary image after applying some simple enhancing methods to partially overcome time consuming problem of the Gabor filter.
Enhanced Algorithm for Obstacle Detection and Avoidance Using a Hybrid of Pla...IOSR Journals
This document presents an enhanced algorithm for obstacle detection and avoidance using a hybrid of plane-to-plane homography, image segmentation, corner detection, and edge detection techniques. The algorithm aims to improve upon previous methods by eliminating false positives, reducing unreliable corners and broken edges, providing depth perception without planar assumptions, and requiring less processing power. The key components of the algorithm include plane-to-plane homography, image segmentation, Canny edge detection, Harris corner detection, and the RANSAC sampling method for system analysis. Test results on sample images show the algorithm can accurately detect obstacles based on texture differences while reducing noise from ground plane textures.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology
International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity.
Wavelet-Based Warping Technique for Mobile Devicescsandit
The document proposes a wavelet-based warping technique to render novel views of compressed images on mobile devices. It uses Haar wavelet transform to compress large reference and depth images, reducing their size. The technique decomposes the images into approximation and detail parts, but only uses the approximation parts for warping. This improves rendering speed on mobile devices. The framework is implemented using Android tools and experiments show it provides faster rendering times for large images compared to direct warping without compression.
This document summarizes an object tracking algorithm that uses Dual Tree Complex Wavelet Transform (DTCWT) feature vectors. It begins by extracting a position vector for the object in the first frame. Nine position vectors are then calculated by shifting the original position in different directions. The second frame is cropped into nine blocks based on these position vectors. DTCWT is used to extract feature vectors for the block in the first frame and nine blocks in the second frame. Block matching is performed by calculating the Manhattan distance between the first frame's feature vector and each of the second frame's nine vectors. The block with the minimum distance corresponds to the tracked object. This process is repeated over successive frames to track the moving object.
This document discusses single object tracking and velocity determination. It begins with an introduction and objectives of the project which is to develop an algorithm for tracking a single object and determining its velocity in a sequence of video frames. It then provides details on preprocessing techniques like mean filtering, Gaussian smoothing and median filtering to reduce noise. It describes segmentation methods including histogram-based, single Gaussian background and frame difference approaches. Feature extraction methods like edges, bounding boxes and color are explained. Object detection using optical flow and block matching is covered. Finally, it discusses tracking and calculating velocity of the moving object. MATLAB is introduced as a technical computing language for solving these types of problems.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology
Conventional non-vision based navigation systems relying on purely Global Positioning System (GPS) or inertial sensors can provide the 3D position or orientation of the user. However GPS is often not available in forested regions and cannot be used indoors. Visual odometry provides an independent method to estimate position and orientation of the user/system based on the images captured by the moving user accurately. Vision based systems also provide information (e.g. images, 3D location of landmarks, detection of scene objects) about the scene that the user is looking at. In this project, a set of techniques are used for the accurate pose and position estimation of the moving vehicle for autonomous navigation using the images obtained from two cameras placed at two different locations of the same area on the top of the vehicle. These cases are referred to as stereo vision. Stereo vision provides a method for the 3D reconstruction of the environment which is required for pose and position estimation. Firstly, a set of images are captured. The Harris corner detector is utilized to automatically extract a set of feature points from the images and then feature matching is done using correlation based matching. Triangulation is applied on feature points to find the 3D co-ordinates. Next, a new set of images is captured. Then repeat the same technique for the new set of images too. Finally, by using the 3D feature points, obtained from the first set of images and the new set of images, the pose and position estimation of moving vehicle is done using QUEST algorithm.
Gait Based Person Recognition Using Partial Least Squares Selection Scheme ijcisjournal
The document summarizes a research paper on gait-based person recognition using partial least squares selection. It presents an Arbitrary View Transformation Model (AVTM) that uses gait energy images and partial least squares (PLS) feature selection to improve gait recognition accuracy under varying viewing angles, clothing, and other conditions. The proposed AVTM PLS method is evaluated on the CASIA gait database and shown to achieve higher recognition rates compared to other existing methods, especially when there are changes in viewing angle, clothing, or whether the person is carrying something. Tables of results demonstrate the proposed method outperforms alternatives across different test conditions and ranges of gallery and probe viewing angles.
Enhanced target tracking based on mean shift algorithm for satellite imageryeSAT Journals
Abstract Target tracking in high resolution satellite images is challenging task for computer vision field. In this paper we have proposed a mean shift algorithm based enhanced target tracking system for high resolution satellite imagery. In proposed tracking algorithm, Target modeling is done using spectral features of target object i.e. Mean & Energy density function. Feature Vector space with minimum Euclidean Distance is used for predicting next possible position of target object in consecutive frames. Proposed tracking algorithm has been tested using two high resolution databases i.e. Harbor & Airport region database acquired by WorldView-2 satellite at different times. Recall, Precision & F1 score etc. performance parameters are also calculated for showing the tracking ability of the proposed method in real-time applications and are compared with the results of Regional Operator Design based tracking algorithm proposed in [1]. The results show that our proposed method gives relatively better performance than the other tracking algorithms used in satellite imagery. Keywords- Target tracking, Mean shift algorithm, Energy density function, Feature Vector Space, Frame
Design and optimization of ion propulsion dronebjmsejournal
Electric propulsion technology is widely used in many kinds of vehicles in recent years, and aircrafts are no exception. Technically, UAVs are electrically propelled but tend to produce a significant amount of noise and vibrations. Ion propulsion technology for drones is a potential solution to this problem. Ion propulsion technology is proven to be feasible in the earth’s atmosphere. The study presented in this article shows the design of EHD thrusters and power supply for ion propulsion drones along with performance optimization of high-voltage power supply for endurance in earth’s atmosphere.
More Related Content
Similar to Dense Visual Odometry Using Genetic Algorithm
Data-Driven Motion Estimation With Spatial AdaptationCSCJournals
The pel-recursive computation of 2-D optical flow raises a wealth of issues, such as the treatment of outliers, motion discontinuities and occlusion. Our proposed approach deals with these issues within a common framework. It relies on the use of a data-driven technique called Generalised Cross Validation to estimate the best regularisation scheme for a given pixel. In our model, the regularisation parameter is a general matrix whose entries can account for different sources of error. The motion vector estimation takes into consideration local image properties following a spatially adaptive approach where each moving pixel is supposed to have its own regularisation matrix. Preliminary experiments indicate that this approach provides robust estimates of the optical flow.
Stereo Vision Distance Estimation Employing Canny Edge Detector with Interpol...ZaidHussein6
This document summarizes a research paper that proposes a stereo vision algorithm called the Canny Block Matching Algorithm (CBMA) to estimate distance from stereo images. CBMA uses the Canny edge detector to extract edges from images and block matching with Sum of Absolute Difference (SAD) to determine disparity maps and reduce processing time. The algorithm was tested on stereo image pairs and achieved an error reduction of about 2% and processing time reduction compared to other methods. Interpolation techniques including bilinear, 1st order polynomial and 2nd order polynomial were also evaluated to enhance the output images and further reduce errors.
Goal location prediction based on deep learning using RGB-D camerajournalBEEI
In the navigation system, the desired destination position plays an essential role since the path planning algorithms takes a current location and goal location as inputs as well as the map of the surrounding environment. The generated path from path planning algorithm is used to guide a user to his final destination. This paper presents a proposed algorithm based on RGB-D camera to predict the goal coordinates in 2D occupancy grid map for visually impaired people navigation system. In recent years, deep learning methods have been used in many object detection tasks. So, the object detection method based on convolution neural network method is adopted in the proposed algorithm. The measuring distance between the current position of a sensor and the detected object depends on the depth data that is acquired from RGB-D camera. Both of the object detected coordinates and depth data has been integrated to get an accurate goal location in a 2D map. This proposed algorithm has been tested on various real-time scenarios. The experiments results indicate to the effectiveness of the proposed algorithm.
This project is to retrieve the similar geographic images from the dataset based on the features extracted.
Retrieval is the process of collecting the relevant images from the dataset which contains more number of
images. Initially the preprocessing step is performed in order to remove noise occurred in input image with
the help of Gaussian filter. As the second step, Gray Level Co-occurrence Matrix (GLCM), Scale Invariant
Feature Transform (SIFT), and Moment Invariant Feature algorithms are implemented to extract the
features from the images. After this process, the relevant geographic images are retrieved from the dataset
by using Euclidean distance. In this, the dataset consists of totally 40 images. From that the images which
are all related to the input image are retrieved by using Euclidean distance. The approach of SIFT is to
perform reliable recognition, it is important that the feature extracted from the training image be
detectable even under changes in image scale, noise and illumination. The GLCM calculates how often a
pixel with gray level value occurs. While the focus is on image retrieval, our project is effectively used in
the applications such as detection and classification.
This document proposes a method for change detection in images that combines Change Vector Analysis, K-Means clustering, Otsu thresholding, and mathematical morphology. It involves detecting intensity changes using CVA, segmenting the difference image using K-Means, calculating a threshold with Otsu's method, applying the threshold and morphological operations, and comparing results to other change detection techniques. Experimental results on medical and other images show the proposed method achieves satisfactory change detection with fewer errors compared to other methods.
3D Reconstruction from Multiple uncalibrated 2D Images of an ObjectAnkur Tyagi
3D reconstruction is the process of capturing the shape and appearance of real objects. In this project we are using passive methods which only use sensors to measure the radiance reflected or emitted by the objects surface to infer its 3D structure.
This document proposes using dual back-to-back Kinect sensors mounted on a robot to capture a 3D model of a large indoor scene. Traditionally, one Kinect is slid across an area, but this requires prominent features and careful handling. The dual Kinect setup requires calibrating the relative pose between the sensors. Since they do not share a view, traditional calibration is not possible. The authors place a dual-face checkerboard on top with a mirror to enable each Kinect to view the same calibration object. This allows estimating the pose between the sensors using a mirror-based algorithm. After capturing local models, the two Kinect views can be merged into a combined 3D model with a larger field of view.
This document describes a laser distance measurement system using a webcam. It consists of a laser transmitter and webcam receiver. The laser pulse is reflected off an object and received by the webcam. Software calculates the distance based on the time of flight. The system achieves high accuracy of ±3cm. It calibrates the system using test measurements to determine the relationship between pixel location of the laser dot and actual distance. This allows accurate distance measurements within a few percent of error out to over 2 meters. Potential improvements discussed are using a laser line instead of dot for more data points and a green laser for better visibility.
Gabor filter is a powerful way to enhance biometric images like fingerprint images in order to extract correct features from these images, Gabor filter used in extracting features directly asin iris images, and sometimes Gabor filter has been used for texture analysis. In fingerprint images The even symmetric Gabor filter is contextual filter or multi-resolution filter will be used to enhance fingerprint imageby filling small gaps (low-pass effect) in the direction of the ridge (black regions) and to increase the discrimination between ridge and valley (black and white regions) in the direction, orthogonal to the ridge, the proposed method in applying Gabor filter on fingerprint images depending on translated fingerprint image into binary image after applying some simple enhancing methods to partially overcome time consuming problem of the Gabor filter.
Enhanced Algorithm for Obstacle Detection and Avoidance Using a Hybrid of Pla...IOSR Journals
This document presents an enhanced algorithm for obstacle detection and avoidance using a hybrid of plane-to-plane homography, image segmentation, corner detection, and edge detection techniques. The algorithm aims to improve upon previous methods by eliminating false positives, reducing unreliable corners and broken edges, providing depth perception without planar assumptions, and requiring less processing power. The key components of the algorithm include plane-to-plane homography, image segmentation, Canny edge detection, Harris corner detection, and the RANSAC sampling method for system analysis. Test results on sample images show the algorithm can accurately detect obstacles based on texture differences while reducing noise from ground plane textures.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology
International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity.
Wavelet-Based Warping Technique for Mobile Devicescsandit
The document proposes a wavelet-based warping technique to render novel views of compressed images on mobile devices. It uses Haar wavelet transform to compress large reference and depth images, reducing their size. The technique decomposes the images into approximation and detail parts, but only uses the approximation parts for warping. This improves rendering speed on mobile devices. The framework is implemented using Android tools and experiments show it provides faster rendering times for large images compared to direct warping without compression.
This document summarizes an object tracking algorithm that uses Dual Tree Complex Wavelet Transform (DTCWT) feature vectors. It begins by extracting a position vector for the object in the first frame. Nine position vectors are then calculated by shifting the original position in different directions. The second frame is cropped into nine blocks based on these position vectors. DTCWT is used to extract feature vectors for the block in the first frame and nine blocks in the second frame. Block matching is performed by calculating the Manhattan distance between the first frame's feature vector and each of the second frame's nine vectors. The block with the minimum distance corresponds to the tracked object. This process is repeated over successive frames to track the moving object.
This document discusses single object tracking and velocity determination. It begins with an introduction and objectives of the project which is to develop an algorithm for tracking a single object and determining its velocity in a sequence of video frames. It then provides details on preprocessing techniques like mean filtering, Gaussian smoothing and median filtering to reduce noise. It describes segmentation methods including histogram-based, single Gaussian background and frame difference approaches. Feature extraction methods like edges, bounding boxes and color are explained. Object detection using optical flow and block matching is covered. Finally, it discusses tracking and calculating velocity of the moving object. MATLAB is introduced as a technical computing language for solving these types of problems.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology
Conventional non-vision based navigation systems relying on purely Global Positioning System (GPS) or inertial sensors can provide the 3D position or orientation of the user. However GPS is often not available in forested regions and cannot be used indoors. Visual odometry provides an independent method to estimate position and orientation of the user/system based on the images captured by the moving user accurately. Vision based systems also provide information (e.g. images, 3D location of landmarks, detection of scene objects) about the scene that the user is looking at. In this project, a set of techniques are used for the accurate pose and position estimation of the moving vehicle for autonomous navigation using the images obtained from two cameras placed at two different locations of the same area on the top of the vehicle. These cases are referred to as stereo vision. Stereo vision provides a method for the 3D reconstruction of the environment which is required for pose and position estimation. Firstly, a set of images are captured. The Harris corner detector is utilized to automatically extract a set of feature points from the images and then feature matching is done using correlation based matching. Triangulation is applied on feature points to find the 3D co-ordinates. Next, a new set of images is captured. Then repeat the same technique for the new set of images too. Finally, by using the 3D feature points, obtained from the first set of images and the new set of images, the pose and position estimation of moving vehicle is done using QUEST algorithm.
Gait Based Person Recognition Using Partial Least Squares Selection Scheme ijcisjournal
The document summarizes a research paper on gait-based person recognition using partial least squares selection. It presents an Arbitrary View Transformation Model (AVTM) that uses gait energy images and partial least squares (PLS) feature selection to improve gait recognition accuracy under varying viewing angles, clothing, and other conditions. The proposed AVTM PLS method is evaluated on the CASIA gait database and shown to achieve higher recognition rates compared to other existing methods, especially when there are changes in viewing angle, clothing, or whether the person is carrying something. Tables of results demonstrate the proposed method outperforms alternatives across different test conditions and ranges of gallery and probe viewing angles.
Enhanced target tracking based on mean shift algorithm for satellite imageryeSAT Journals
Abstract Target tracking in high resolution satellite images is challenging task for computer vision field. In this paper we have proposed a mean shift algorithm based enhanced target tracking system for high resolution satellite imagery. In proposed tracking algorithm, Target modeling is done using spectral features of target object i.e. Mean & Energy density function. Feature Vector space with minimum Euclidean Distance is used for predicting next possible position of target object in consecutive frames. Proposed tracking algorithm has been tested using two high resolution databases i.e. Harbor & Airport region database acquired by WorldView-2 satellite at different times. Recall, Precision & F1 score etc. performance parameters are also calculated for showing the tracking ability of the proposed method in real-time applications and are compared with the results of Regional Operator Design based tracking algorithm proposed in [1]. The results show that our proposed method gives relatively better performance than the other tracking algorithms used in satellite imagery. Keywords- Target tracking, Mean shift algorithm, Energy density function, Feature Vector Space, Frame
Similar to Dense Visual Odometry Using Genetic Algorithm (20)
Design and optimization of ion propulsion dronebjmsejournal
Electric propulsion technology is widely used in many kinds of vehicles in recent years, and aircrafts are no exception. Technically, UAVs are electrically propelled but tend to produce a significant amount of noise and vibrations. Ion propulsion technology for drones is a potential solution to this problem. Ion propulsion technology is proven to be feasible in the earth’s atmosphere. The study presented in this article shows the design of EHD thrusters and power supply for ion propulsion drones along with performance optimization of high-voltage power supply for endurance in earth’s atmosphere.
Discover the latest insights on Data Driven Maintenance with our comprehensive webinar presentation. Learn about traditional maintenance challenges, the right approach to utilizing data, and the benefits of adopting a Data Driven Maintenance strategy. Explore real-world examples, industry best practices, and innovative solutions like FMECA and the D3M model. This presentation, led by expert Jules Oudmans, is essential for asset owners looking to optimize their maintenance processes and leverage digital technologies for improved efficiency and performance. Download now to stay ahead in the evolving maintenance landscape.
Software Engineering and Project Management - Introduction, Modeling Concepts...Prakhyath Rai
Introduction, Modeling Concepts and Class Modeling: What is Object orientation? What is OO development? OO Themes; Evidence for usefulness of OO development; OO modeling history. Modeling
as Design technique: Modeling, abstraction, The Three models. Class Modeling: Object and Class Concept, Link and associations concepts, Generalization and Inheritance, A sample class model, Navigation of class models, and UML diagrams
Building the Analysis Models: Requirement Analysis, Analysis Model Approaches, Data modeling Concepts, Object Oriented Analysis, Scenario-Based Modeling, Flow-Oriented Modeling, class Based Modeling, Creating a Behavioral Model.
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...PriyankaKilaniya
Energy efficiency has been important since the latter part of the last century. The main object of this survey is to determine the energy efficiency knowledge among consumers. Two separate districts in Bangladesh are selected to conduct the survey on households and showrooms about the energy and seller also. The survey uses the data to find some regression equations from which it is easy to predict energy efficiency knowledge. The data is analyzed and calculated based on five important criteria. The initial target was to find some factors that help predict a person's energy efficiency knowledge. From the survey, it is found that the energy efficiency awareness among the people of our country is very low. Relationships between household energy use behaviors are estimated using a unique dataset of about 40 households and 20 showrooms in Bangladesh's Chapainawabganj and Bagerhat districts. Knowledge of energy consumption and energy efficiency technology options is found to be associated with household use of energy conservation practices. Household characteristics also influence household energy use behavior. Younger household cohorts are more likely to adopt energy-efficient technologies and energy conservation practices and place primary importance on energy saving for environmental reasons. Education also influences attitudes toward energy conservation in Bangladesh. Low-education households indicate they primarily save electricity for the environment while high-education households indicate they are motivated by environmental concerns.
Electric vehicle and photovoltaic advanced roles in enhancing the financial p...IJECEIAES
Climate change's impact on the planet forced the United Nations and governments to promote green energies and electric transportation. The deployments of photovoltaic (PV) and electric vehicle (EV) systems gained stronger momentum due to their numerous advantages over fossil fuel types. The advantages go beyond sustainability to reach financial support and stability. The work in this paper introduces the hybrid system between PV and EV to support industrial and commercial plants. This paper covers the theoretical framework of the proposed hybrid system including the required equation to complete the cost analysis when PV and EV are present. In addition, the proposed design diagram which sets the priorities and requirements of the system is presented. The proposed approach allows setup to advance their power stability, especially during power outages. The presented information supports researchers and plant owners to complete the necessary analysis while promoting the deployment of clean energy. The result of a case study that represents a dairy milk farmer supports the theoretical works and highlights its advanced benefits to existing plants. The short return on investment of the proposed approach supports the paper's novelty approach for the sustainable electrical system. In addition, the proposed system allows for an isolated power setup without the need for a transmission line which enhances the safety of the electrical network
Build the Next Generation of Apps with the Einstein 1 Platform.
Rejoignez Philippe Ozil pour une session de workshops qui vous guidera à travers les détails de la plateforme Einstein 1, l'importance des données pour la création d'applications d'intelligence artificielle et les différents outils et technologies que Salesforce propose pour vous apporter tous les bénéfices de l'IA.
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODELijaia
As digital technology becomes more deeply embedded in power systems, protecting the communication
networks of Smart Grids (SG) has emerged as a critical concern. Distributed Network Protocol 3 (DNP3)
represents a multi-tiered application layer protocol extensively utilized in Supervisory Control and Data
Acquisition (SCADA)-based smart grids to facilitate real-time data gathering and control functionalities.
Robust Intrusion Detection Systems (IDS) are necessary for early threat detection and mitigation because
of the interconnection of these networks, which makes them vulnerable to a variety of cyberattacks. To
solve this issue, this paper develops a hybrid Deep Learning (DL) model specifically designed for intrusion
detection in smart grids. The proposed approach is a combination of the Convolutional Neural Network
(CNN) and the Long-Short-Term Memory algorithms (LSTM). We employed a recent intrusion detection
dataset (DNP3), which focuses on unauthorized commands and Denial of Service (DoS) cyberattacks, to
train and test our model. The results of our experiments show that our CNN-LSTM method is much better
at finding smart grid intrusions than other deep learning algorithms used for classification. In addition,
our proposed approach improves accuracy, precision, recall, and F1 score, achieving a high detection
accuracy rate of 99.50%.
Redefining brain tumor segmentation: a cutting-edge convolutional neural netw...IJECEIAES
Medical image analysis has witnessed significant advancements with deep learning techniques. In the domain of brain tumor segmentation, the ability to
precisely delineate tumor boundaries from magnetic resonance imaging (MRI)
scans holds profound implications for diagnosis. This study presents an ensemble convolutional neural network (CNN) with transfer learning, integrating
the state-of-the-art Deeplabv3+ architecture with the ResNet18 backbone. The
model is rigorously trained and evaluated, exhibiting remarkable performance
metrics, including an impressive global accuracy of 99.286%, a high-class accuracy of 82.191%, a mean intersection over union (IoU) of 79.900%, a weighted
IoU of 98.620%, and a Boundary F1 (BF) score of 83.303%. Notably, a detailed comparative analysis with existing methods showcases the superiority of
our proposed model. These findings underscore the model’s competence in precise brain tumor localization, underscoring its potential to revolutionize medical
image analysis and enhance healthcare outcomes. This research paves the way
for future exploration and optimization of advanced CNN models in medical
imaging, emphasizing addressing false positives and resource efficiency.
VARIABLE FREQUENCY DRIVE. VFDs are widely used in industrial applications for...PIMR BHOPAL
Variable frequency drive .A Variable Frequency Drive (VFD) is an electronic device used to control the speed and torque of an electric motor by varying the frequency and voltage of its power supply. VFDs are widely used in industrial applications for motor control, providing significant energy savings and precise motor operation.
2008 BUILDING CONSTRUCTION Illustrated - Ching Chapter 02 The Building.pdf
Dense Visual Odometry Using Genetic Algorithm
1. International Journal of
INTELLIGENT SYSTEMS AND APPLICATIONS IN
ENGINEERING
ISSN:2147-67992147-6799 www.ijisae.org Original Research Paper
International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 611
Dense Visual Odometry Using Genetic Algorithm
Slimane Djema*1
, Zoubir Abdeslem Benselama2
, Ramdane Hedjar3
, Krabi Abdallah4
Submitted: 25/04/2023 Revised: 29/06/2023 Accepted: 08/07/2023
Abstract: Our work aims to estimate the camera motion mounted on the head of a mobile robot or a moving object from RGB-D images
in a static scene. The problem of motion estimation is transformed into a nonlinear least squares function. Methods for solving such
problems are iterative. Various classic methods gave an iterative solution by linearizing this function. We can also use the metaheuristic
optimization method to solve this problem and improve results. In this paper, a new algorithm is developed for visual odometry using a
sequence of RGB-D images. This algorithm is based on a genetic algorithm. The proposed iterative genetic algorithm searches using
particles to estimate the optimal motion and then compares it to the traditional methods. To evaluate our method, we use the root mean
square error to compare it with the based energy method and another metaheuristic method. We prove the efficiency of our innovative
algorithm on a large set of images.
Keywords: camera motion, genetic Algorithm, RGB-D images, static scene, visual odometry
1. Introduction
In navigation, odometry is the use of information coming
from sensors such as rotary encoders placed in actuators to
estimate the position of a moving vehicle. This method has
limited use because the wheels can slide on smooth surfaces
or spin motionless in the sand, causing an error in the
calculation of the movement. This error will accumulate
over time and create an increased divergence between the
real and estimated positions during the motion.
Visual odometry is the operation of estimating the position
using sequential frames coming from an onboard camera
analyzed by different algorithms. Visual odometry enhances
navigation accuracy in mobile objects using any type of
locomotion, such as robots with legs or wheels, on a sticky
or sandy surface or drones. This operation improves the
performance of the robot for the tasks assigned to it.
In our case, we exploited the dense method, which uses the
entire RGB-D image provided by a Microsoft Kinect sensor.
RGB-D cameras refer to digital cameras that provide the
colors red, green and blue (RGB) and depth (D) as
information for every pixel in the image. The website [1]
gives RGBD images and provide associated information as
truth trajectory and camera calibration.
In this work, we propose an algorithm for visual odometry
that uses a genetic algorithm (GA). This algorithm is
presented in Section 3, where we will talk first about
modeling the deduction of motion from the vision by a
minimization error equation, and then we will explain how
to construct the warping image that is used to calculate the
error equation. After that, we will talk in detail about the
genetic algorithm. In section 4, we present how we compute
the relative pose error used to evaluate methods, and in this
section, the accuracy of our algorithm is validated with a
large set of experiments. The conclusion of our work is
given in Section 5.
2. Related Work
Visual odometry calculates the object motion using images
captured while the robot is moving. Diffirent approaches for
visual odometry were distinguished, Among them dense and
sparse approaches.
The sparse approach uses feature extraction from the image
as noted in [2], [3] or [4] and [5].
A dense approach uses the entire information in the image
for motion estimation as in [6]. The first dense approach was
presented by [7] which uses image alignment and
minimization of geometrical error distance as described in
[8] and [9].
Lately, after the discovery of the RGB-D image, the uses of
this format have become wide in visual odometry, as given
by [10], [11], [12] or [13].
In the literature, there are many optimization methods using
metaheuristic approaches explored for the problem of
motion estimation by vision using sparse methods such as
[14], [15], [16], in addition [17], and [18], [19], or dense
methods as in [20] and [21].
In our approach, we estimate the motion by a dense visual
odometry method using a metaheuristic algorithm and
RGBD images as a data set.
1
Department of Electronics, Saad Dahlab University, Blida, Algeria
2
Department of Electronics, Saad Dahlab University, Blida, Algeria
3
Department of Computer Engineering, King Saud University, Riyadh,
Saudi Arabia
4
Department of Electronics, Saad Dahlab University, Blida, Algeria
* Corresponding Author Email: djema20132@gmail.com
2. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 612
3. Method
In this part, we explain in a simplified and integrated way the
framework for motion estimation between two RGB-D
images in a static scene as described in [8] using a genetic
algorithm.
3.1. System modeling
Visual odometry aims to calculate the motion estimated
between two successive frames (It , It+1) captured by a
camera mounted on the head of a mobile robot or a moving
object. We will show how this problem has been modeled as
a mathematical equation.
An image r called residue is calculated by subtracting the
pixel intensity photometric, i.e., the pixel color code average,
of each two pixels at the same position from these two
images. The pixels in r are more enlightened, engendered a
higher error. The following function, as described in [10],
represents the intensity error between two consecutive RGB
frames of N pixels:
E(ξ)=
1
𝑁
∑𝑁
𝑖= 1 |It+1(ω(ξ,pi))-It|²=
1
𝑁
∑𝑁
𝑖=1 |ri(ξ)|²
(1)
Where ξ ϵ ℝ 6
is the motion that we seek to find, and ω(ξ,
pi) is the warping function.
The motion estimation of the camera consists of minimizing
the error of the intensities (also called photometric error) of
all pixels of the image. Theoretically, if the motion vector ξ
of the camera between the two images is perfectly known,
the error of the intensities on all the pixels is null; but in
reality, this error is never null because of the noise of the
sensor and the changes in the visibility angle of objects, etc.
However, this error remains minimal knowing the true
motion vector between the two images. The objective of
these approaches is to find the camera motion estimated ξ
between two images that minimizes the intensities of all the
pixels of the image residual by minimizing the following
function:
ξ=minξE(ξ)=minξ
1
𝑁
∑𝑁
𝑖=1 |ri (ξ)|².
(2)
This equation is solved by a genetic algorithm in this work.
The warp function is considered an essential part of creating
the residual image ri (ξ).
3.2. Building the Warp function
The warp function ω(ξ, p) in equation (1) changes the
position of pixels in It+1 to build a warped image, which we'll
capture by a camera if we move on the inverse of the motion
ξ. Then we subtract this new image from It and judge if the
motion proposed ξ is optimal by calculating the error
equation (1).
The warp function is composed of a set of transformations
shown in Figure 1 as noted in [10], [22] and [23] . A pixel p
of coordinates (u;v;d) of the image It+1 is projected at a point
M in 3D of coordinates (X;Y;Z) or PM by the transformation
P-1
. Then M is transformed from the landmark attached to
It+1 to a point M in 3D of coordinates (X';Y';Z') or PM' in the
landmark of It+1(ω(ξ,pi)) by the transformation g(ξ) as shown
in the following equation:
PM '=g(ξ)×PM .
(3)
Finally, M' is projected in the image plane of the warped
image It+1(ω(ξ,pi)) by the transformation P. Thus, the
function of warp is written in the following form:
ω(ξ,p)=P(g(ξ)P-1
(p)).
(4)
Fig 1. The warp function consists of a set of
transformations that project each pixel in the image It+1
into warped image It+1(ω(ξ,pi)).
Let P be the transformation that makes it possible to go from
a 3D point to a pixel in the image. Each 3D point of
coordinates (X;Y;Z) in space is bound to its corresponding
2D pixel (u;v;d) by the following equation:
𝑃: ℝ3
→ ℝ2
; (𝑋; 𝑌; 𝑍 ) → (𝑢, 𝑣)
{
𝑢 =
𝑋 × 𝑓𝑥
𝑍
+ 𝑐𝑥.
𝑣 =
𝑌 × 𝑓𝑦
𝑍
+ 𝑐𝑦.
(5)
It is possible to rebuild the 3D point of the scene projected
to a pixel thanks to the RGB image type.
The projection P makes it possible to go from a 3D scene to
a 2D frame. Each 3D point of coordinates (X;Y;Z) in space
is bound to its corresponding 2D pixel (u;v;d) by the
following equation:
𝑃−1
: ℝ2
→ ℝ3
; (𝑢, 𝑣, 𝑑) → (𝑋; 𝑌; 𝑍)
{
𝑋 =
𝑢 − 𝑐𝑥
𝑓𝑥
× 𝑑
𝑌 =
𝑣 − 𝑐𝑦
𝑓𝑦
× 𝑑
Z = d .
(6)
3. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 613
With fx, fy, cx and cy are respectively the focal lengths and
the optical centers of the camera. d is the depth of the pixel
returned by the camera.
The motion of a rigid object in 3D space is perfectly
described by its position relative to a fixed reference at all
times. R and T can be grouped into a single rigid
transformation matrix g as:
g=[
𝑅 𝑇
0 1
] ϵℝ4×4
.
(7)
Where R=[
𝑟11 𝑟12 𝑟13
𝑟21 𝑟22 𝑟23
𝑟31 𝑟32 𝑟33
] is the rotation motion, and
T=[
𝑡𝑥
𝑡𝑦
𝑡𝑧
]is the translation motion.
There is a useful minimalistic representation when, for
example, one seeks to determine the parameters using a
numerical optimization method. Indeed, each transformation
matrix describing the motion of a rigid object has a
representation by the vector ξ using six degrees of freedom:
ξ=[ 𝑣1 𝑣3 𝑣 2 𝑤1 𝑤2 𝑤3].
With v = (v1; v2; v3) the linear velocity and w = (w1;w2;w3)
angular velocity.
We use the Lie algebra given in [24] to represent the matrix
g in the function of the twist coordinates ξ; the
transformation g can be calculated from ξ using the
exponential mapping from Lie Algebra to Lie Group as
described in [10] and [25]:
g(ξ)=eξ
̂
(8)
With 𝜉
̂is the anti-symmetric matrix equal to:
ξ
̂=[
0 − 𝑤3 𝑤2 𝑣1
𝑤3 0 − 𝑤1 𝑣2
−𝑤2 𝑤1 0 𝑣3
0 0 0 0
]
Thus, we talked about all the components of equation (4) as
ξ, g(ξ), P, and P-1
, which build the warping of each pixel in
the image It+1, and so we get It+1(ω(ξ,pi)), then we can
calculate the residual image and photometric error using the
equation (1). As for ξ, several values are proposed by GA
as a motion for the particles, after that we calculate the
corresponding error value for each particle, as we will
explain in our proposed algorithm.
3.3. Genetic Algorithm for motion estimation
Genetic algorithms, a type of optimization algorithm, were
developed in the 1970s to comprehend the reproduction field
of living organisms and the behavior of their genes. Then,
they have been applied as an algorithm in machine learning.
The main role of the genetic algorithm is to generate several
motions based on previously existing motions using their
specified equations, and then we will choose the motions that
result in less error using the function (1), repeat the same
work in the next iterations until the stopping conditions are
reached, and finally determine the best motion ever ξ using
the function (2).
The explanation of the genetic algorithm (GA) design as
mentioned in [26] is the following:
3.3.1. Representation
The position ξ has been identified as a chromosome, and the
decision variables in position ξ are genes with 6 alleles; the
first three are reserved for the linear velocity and the last
three are reserved for the angular velocity.
3.3.2. Population initialization
Each particle of the population must have an initial position
ξ using an array of continuous uniform random numbers,
each variable is limited by a lower and upper bound.
3.3.3. Objective Function
This function is a mathematical equation, and its maximum
value corresponds to the best position estimated in the
scene. The term fitness refers to equation (10).
3.3.4. Selection strategy
At this stage, particles are selected for reproduction, and
several methods can be used for this. In our experience, we
use the roulette wheel selection method, which will give each
particle pi of the population a value of probability probi that
is proportional to its fitness value, as mentioned in equation
(10). We note that Ei is the error of individual pi and Emin is
the minimum error, and in the following, the fitness fi of the
particle pi is
fi=exp(-8*Ei/Emin).
(8)
Its probability of being selected is
probi=fi /(∑ 𝑛
𝑖 = 1 fi) .
(9)
Then we apply the cumulative sum of elements to each
selection probability
probci=cumsum(probi).
(10)
The random choice of µ particles for mating is made through
an indiscriminate variable as an independent spin of the
roulette wheel. Better individuals, or the individuals that
have the minimum error or the maximum fitness, have more
chances to be chosen for the next stage thanks to equation
(10).
3.3.5. Reproduction strategy
In our implementation code in Matlab, this strategy consists
of two processes to create new particles.
• Mutation: this process is done on each particle separately.
Thirty percent of the population are selected at random to
4. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 614
undergo the mutation. The mutation rate (pm=0.1) is the
probability of a change in particle motion selected for
mutation.
The formula for mutation is:
ξ′=ξ+M.
(11)
and M is a random vector of mutation computed from:
M=δi*randn(size_ξ)
(12)
where δi is calculated by the formula:
δi= pm *(ξi
U
-ξi
L
)
(13)
where ξi
U
(resp.ξi
L
) represents the upper bound (resp. lower
bound) for ξi.
• Crossover: the resulting parents using the roulette wheel
selection method will be used in recombination (The
crossover process). The goal of recombination is to create
offspring that carry genes from both parents.
Among the crossover processes most commonly used are the
intermediate crossover processes. The intermediate
crossover attempts to average the positions corresponding to
the two parents. The equations of crossover create two
individuals O1 and O2 using the weighted average:
{
𝑂1𝑖 = 𝛼𝜉1𝑖 + (1– 𝛼)𝜉2𝑖
𝑂2𝑖 = 𝛼𝜉2𝑖 + (1– 𝛼)𝜉1𝑖
(14)
The extra range factor for crossover α represents the
proportion of parent choice as random arrays from the
continuous uniform distribution. Then, the new and old
particles will create the future population.
3.3.6. Replacement strategy
The new offspring and the old particles compete for
existence in the future population. We do this by creating a
merged population made up of the previous elements and the
offspring resulting from crossover and mutation strategies,
then we sort order the population using errors calculated by
the function (1) and choosing the ones that have the
minimum errors according to the required number of
particles. Then updating the minimum error ever found.
3.4. Pyramid Multi-resolution
Equation (2) is solved by minimizing the intensities of all
the pixels of the image residual, and its solution is closer to
the truth for tiny motion ξ or small image resolution. To
improve the final motion estimation, we present a pyramid
as mentioned in [22], [27] and [28], where the down-
sampled resolution (DSR) of each image is performed by a
factor of 2 (see Figure 2). In the first, we calculate ξ with the
image corresponding to high-level DSR (level 4) and this
motion will be used as initialization for the next low level in
the pyramid up to the initial resolution of the image, where
we deduce the optimal motion ξ.
Fig 2. Visual representation of an image pyramid with 5
levels.
3.5. Stopping criteria
There are a lot of stopping criteria during the execution of
code metaheuristics. We used two procedures for stopping
below:
• Static procedure: the stopping of the execution must
be known a priori using a maximum value of iterations
for each DSR level of image resolution.
• Diversity procedure: the end of the execution of code
must be when the best particle error stagnates within a
specified number of iterations, keeping the execution
of the algorithm useless.
3.6. Overall algorithm
In the following, we present the most important step that GA
goes through to reach the best motion.
1. Initialization
a. Set the number of particles as N.
b. Set the number of GA iterations as M.
c. Set the variables bound
d. Set crossover and mutation percentage
e. For i=1,…,N, set ξ0
i
=rand(1).
f. Set initial parameters camera intrinsic
2. Main Loop
for j=1:M iterations.
-Select parents using the roulette wheel selection
for i=1:ns particles selected (also called Parents)
-Update the particles via Crossover and Mutation
end
5. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 615
-Evaluate f(Pj
i
) and update Pi
and get ξ of Pbest
end
Retenuξ
The main role of this algorithm is to generate several
motions and calculate the errors produced by equation (1),
then deduce the best motion that corresponds to the
minimum error. After that, it repeats this operation several
iterations until it reaches an optimal motion.
First, this algorithm gives initial motion ξ for each item in
the group of particles, and then a specific number of
particles are selected and used to generate new particles.
After that, these new particles replace the old ones, which
corresponds to a greater value of error E(ξ). Thus, it has
completed one iteration. Then it selects a new group of
particles for the next iteration, and follows the same
previous steps. After the stopping criteria are met, the
motion of the best particle is determined.
We will evaluate this algorithm with several experiments.
4. Evaluation
In this section, we evaluate our method for motion
estimation on a static scene using RGB-D frames that are
available on [1]. For this, we use RPE, RMSE and 3D
trajectory to compare our method to particle swarm
optimization (PSO), and energy-based as a classic method,
which are mentioned respectively in [20] and [12].
4.1. Dataset
We get the RGB-D datasets existing on the website [1] from
the Kinect camera.
(a)
(b)
Fig 3. RGB (a) and depth (b) image from “fr2/desk”
sequence.
Figure 3(a) represents an RGB frame and Figure 3(b) the
corresponding depth, using these two images we have data
of 3D scene that formulate an RGB-D image.
4.2. Relative pose error
Relative Pose Error (RPE) calculate the drift of the
trajectory estimated to the truth trajectory as described in
[29], [30] and [31] in time interval ∆ at step i as
Ei=(Qi
− 1
Qi+∆)−1
(Pi
−1
Pi+∆).
(15)
From a sequence of n images, m=n−∆ is individual relative
pose errors. We define the root mean squared error (RMSE)
as
RMSE(E1:n ,∆)=(
1
𝑚
∑ ||𝑡𝑟𝑎𝑛𝑠(𝐸𝑖)||²
𝑚
𝑖=1 )1/2
.
(16)
Where trans(Ei) is the translational component of RPE.
4.3. Real-time graphical user interfaces
Multicriteria optimization is the selection of the best
element from some set of available alternatives. The initial
parameters of a metaheuristic method like particles number,
probability value, and others are considered criteria and we
must manipulate them to achieve optimal results, this is
what prompted us to build a graphical user interfaces (GUI)
in MATLAB shown in Figure4 for various aspects of
execution code result in real time to follow the results and
choose the best value of criteria to get an optimal solution.
6. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 616
Fig 4. The real-time graphical user interfaces view code
execution in MATLAB
The graphical user interfaces represent in figure 4 contains
four windows, that’s On the right there is a three-
dimensional space in the form of a parallelogram with
dimensions representing the field of motion of the particles,
showing us how the particles behave in their search for the
value of motion by searching for the lowest error value in
equation (1), and in the center of the finite space we notice
a green disk representing the three-dimensional position In
which the previous image was taken, and the red star
represents the location where the next image was taken, and
it is the place where the colored particles are looking for, as
their color reflects the error value, as whenever the lowest
error value or BestCost is in blue, and this allows us to
observe the behavior of particles and evaluating the
proposed method. As for the red arrow, it represents the
direction of the error overlapping with the sequence of
images with a three-dimensional ray, and its length is
mentioned above. In addition, the number of particles is
mentioned and the Down Sampled Resolution DSR and
RMSE correspond to the best particle, and this image clearly
shows the behavior of the particles in finding the motion
between two consecutive photos. The final RMSE between
two consecutive images corresponding to the best particle is
represented in the form of impulses as shown in the upper
graphical example, and on the right is a graph corresponding
to camera trajectory RMSE, either at the bottom, a graph
represents the truth and estimated trajectory in the same 3D
scene.
4.4. Experimental setup
We evaluated our algorithm through various experiments in
a static environment using RGB-D images of dimension
640×480 with a frame rate of 30Hz. These images and their
corresponding ground truths are available on the website
[1].
In our first experiment, we used 90 consecutive frames of
rgbd_dataset_freiburg1_xyz. The function (17) gives the
distance error between estimated motion and ground truth.
Therefore, using this function, we compute the camera
trajectory error of different methods through 90 consecutive
images, and we represent these results in the same graph in
figure 5. The evolution of distance error indicates that the
accumulative error of 90 frames related to the classic
method is the least, but the quasi-stabilization of the error of
the PSO method in the last 60 frames allowed it to
outperform RMSE as we can see in table 1, which is
considered the most important evaluation criterion in visual
odometry. Although the final accumulation error of GA is
greater, the error almost preserved its value between the
beginning and the end of the last thirty frames, which helped
him improve the value of translational components of RPE
and thus outperform the RMSE of classic method.
Fig 5. Camera trajectory error of GA, PSO and classic
method using a part of fr1_xyz dataset.
The representation in the 3D scene of the truth and estimated
camera trajectory clearly shows the effectiveness of the
motion estimation methods. Figures 6, 7 and 8 show the
camera trajectory for fr1_xyz of ground truth and different
motion estimation methods; classic and metaheuristics
methods.
Fig 6. Truth camera trajectory and classic method using a
part of fr1_xyz dataset.
Through ninety RGBD frames and in a back and forth path,
the three methods gave very acceptable results, and this
confirms that the error of a representation in Figure 5 is very
small compared to the path traveled, and there was no
deviation away from the true trajectory.
7. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 617
Fig 7. Truth camera trajectory with GA method using a
part of fr1_xyz dataset.
Fig 8. Truth camera trajectory with PSO method using a
part of fr1_xyz dataset.
In the second experiment, we used 60 consecutive frames of
rgbd_dataset_freiburg2_desk. The distance error between
the ground truth and the trajectory of GA, PSO and the
classic method is represented in Figure 9. We notice that in
the first thirty frames, the GA method achieved the least
distance error compared to the other methods, and the
results were close compared to the classic method. This is
clearly shown in the Table I with the superiority of the GA.
Fig 9. Camera trajectory error of GA, PSO and Classic
method using a part of fr2_desk dataset.
The representation in the same 3D scene of the truth and the
estimated camera trajectory clearly shows the effectiveness
of the motion estimation method. Figure 10 shows the
camera trajectory for fr2_desk of ground truth and different
motion estimation methods; classic and meta-heuristics
methods. The three methods gave acceptable results, where
we notice the corresponding trajectories are very close to the
true trajectory, but for the trajectory corresponding to GA,
we notice that it is the closest to the truth trajectory, which
confirms through this experiment that this innovative
method competes with the previous methods; classic and
PSO or even better, and this confirms previously obtained
results from Figure 10 and Table 1.
Fig 10. Truth camera trajectory with GA, PSO and Classic
method using a part of fr2_desk dataset.
Table 1 shows the root mean square error (RMSE)
calculated using the function (18) for two previous methods
GA, PSO and the results of the methods classic tested in
[10].
Table 1. Root mean square error (RMSE) of drift in meters
per second for different methods for ground truth.
Dataset GA PSO Classic
fr1_xyz 0.04062 m 0.03598 m 0.04827 m
fr2_desk 0.01856 m 0.02836 m 0.02524 m
For the dataset (freiburg1_xyz) our method using GA has
proven its efficacy in comparison to the classic method.
Regarding the dataset (freiburg2_desk), our innovative
method using GA has successfully proven its efficiency
compared to both methods used in this experiment: PSO and
the classic method.
5. Conclusion and Future Work
We built a new algorithm to estimate the trajectory of a
camera in a static scene using a metaheuristic method. This
method is a genetic algorithm. After a large set of
experiments, we have demonstrated the efficacy of this
method and the improvement in results. Gradually, we
showed how to achieve the desired results. In the future, we
want to extend this work to estimate body motion in a
dynamic scene.
8. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 618
References
[1] “RGB-D SLAM Dataset and Benchmark,” TUM
Department of Informatics, Technical University of
Munich, Germany, 2011. Accessed: Jan. 01, 2021.
[Online]. Available:
http://vision.in.tum.de/data/datasets/rgbd-dataset.
[2] F. Cheng, C. Liu, H. Wu, and M. Ai, “DIRECT
SPARSE VISUAL ODOMETRY WITH
STRUCTURAL REGULARITIES FOR LONG
CORRIDOR ENVIRONMENTS,” Int. Arch.
Photogramm. Remote Sens. Spatial Inf. Sci., vol.
XLIII-B2-2020, pp. 757–763, Aug. 2020, doi:
10.5194/isprs-archives-XLIII-B2-2020-757-2020.
[3] S. T. Khawase, S. D. Kamble, N. V. Thakur, and A. S.
Patharkar, “An Overview of Block Matching
Algorithms for Motion Vector Estimation,”
Gopeshwar, Uttrakhand, India, Jun. 2017, pp. 217–
222. doi: 10.15439/2017R85.
[4] M. Ghaffari, W. Clark, A. Bloch, R. M. Eustice, and J.
W. Grizzle, “Continuous Direct Sparse Visual
Odometry from RGB-D Images,” . In Proceedings of
the Robotics: Science and Systems Conference,
Freiburg, Germany, Jun. 2019, doi:
10.48550/ARXIV.1904.02266.
[5] N. Zhang and Y. Zhao, “Fast and Robust Monocular
Visua-Inertial Odometry Using Points and Lines,”
Sensors, vol. 19, no. 20, p. 4545, Oct. 2019, doi:
10.3390/s19204545.
[6] B. Canovas, M. Rombaut, A. Negre, D. Pellerin, and
S. Olympieff, “Speed and Memory Efficient Dense
RGB-D SLAM in Dynamic Scenes,” in 2020
IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS), Las Vegas, NV, USA,
Oct. 2020, pp. 4996–5001. doi:
10.1109/IROS45743.2020.9341542.
[7] A. I. Comport, E. Malis, and P. Rives, “Real-time
Quadrifocal Visual Odometry,” The International
Journal of Robotics Research, vol. 29, no. 2–3, pp.
245–266, Feb. 2010, doi:
10.1177/0278364909356601.
[8] B. D. Lucas and T. Kanade, “An Iterative Image
Registration Technique with an Application to Stereo
Vision,” Proceedings of the 7th International Joint
Conference on Artificial Intelligence (IJCAI-81),
Vancouver, BC, Canada, pp. 674–679, Aug. 1981.
[9] P. J. Besl and N. D. McKay, “Method for registration
of 3-D shapes,” in Sensor Fusion IV: Control
Paradigms and Data Structures, Boston, MA, USA,
Apr. 1992, vol. 1611, pp. 586–606. doi:
10.1117/12.57955.
[10] A. Dib and F. Charpillet, “Robust dense visual
odometry for RGB-D cameras in a dynamic
environment,” in 2015 International Conference on
Advanced Robotics (ICAR), Istanbul, Turkey, Jul.
2015, pp. 1–7. doi: 10.1109/ICAR.2015.7298210.
[11] T. Whelan, H. Johannsson, M. Kaess, J. J. Leonard,
and J. McDonald, “Robust real-time visual odometry
for dense RGB-D mapping,” in 2013 IEEE
International Conference on Robotics and
Automation, Karlsruhe, Germany, May 2013, pp.
5724–5731. doi: 10.1109/ICRA.2013.6631400.
[12] F. Steinbrucker, J. Sturm, and D. Cremers, “Real-time
visual odometry from dense RGB-D images,” in 2011
IEEE International Conference on Computer Vision
Workshops (ICCV Workshops), Barcelona, Spain,
Nov. 2011, pp. 719–722. doi:
10.1109/ICCVW.2011.6130321.
[13] C. Kerl, J. Sturm, and D. Cremers, “Robust odometry
estimation for RGB-D cameras,” in 2013 IEEE
International Conference on Robotics and
Automation, Karlsruhe, Germany, May 2013, pp.
3748–3754. doi: 10.1109/ICRA.2013.6631104.
[14] E. Cuevas, D. Zaldívar, M. Pérez-Cisneros, H. Sossa,
and V. Osuna, “Block matching algorithm for motion
estimation based on Artificial Bee Colony (ABC),”
Applied Soft Computing, vol. 13, no. 6, pp. 3047–3059,
Jun. 2013, doi: 10.1016/j.asoc.2012.09.020.
[15] M. Shahbazi, G. Sohn, J. Théau, and P. Ménard,
“ROBUST SPARSE MATCHING AND MOTION
ESTIMATION USING GENETIC ALGORITHMS,”
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.,
vol. XL-3/W2, pp. 197–204, Mar. 2015, doi:
10.5194/isprsarchives-XL-3-W2-197-2015.
[16] M. Tagliasacchi, “A genetic algorithm for optical flow
estimation,” Image and Vision Computing, vol. 25, no.
2, pp. 141–147, Feb. 2007, doi:
10.1016/j.imavis.2006.01.021.
[17] A. Sehgal, A. Singandhupe, H. M. La, A. Tavakkoli,
and S. J. Louis, “Lidar-Monocular Visual Odometry
with Genetic Algorithm for Parameter Optimization,”
in Advances in Visual Computing, Springer
International Publishing., vol. 11845, Cham, Suisse:
Springer International Publishing, 2019, pp. 358–370.
Accessed: Sep. 11, 2022. [Online]. Available:
http://link.springer.com/10.1007/978-3-030-33723-
0_29
[18] Y. K. Yu, K. H. Wong, and M. M. Y. Chang, “Pose
Estimation for Augmented Reality Applications Using
Genetic Algorithm,” IEEE Trans. Syst., Man, Cybern.
B, vol. 35, no. 6, pp. 1295–1301, Dec. 2005, doi:
10.1109/TSMCB.2005.850164.
[19] C.-F. Chao, M.-H. Horng, and Y.-C. Chen, “Motion
Estimation Using the Firefly Algorithm in Ultrasonic
Image Sequence of Soft Tissue,” Computational and
Mathematical Methods in Medicine, vol. 2015, pp. 1–
8, 2015, doi: 10.1155/2015/343217.
[20] Y. K. Baik, J. Kwon, H. S. Lee, and K. M. Lee,
“Geometric particle swarm optimization for robust
9. International Journal of Intelligent Systems and Applications in Engineering IJISAE, 2023, 11(3), 611–619 | 619
visual ego-motion estimation via particle filtering,”
Image and Vision Computing, vol. 31, no. 8, pp. 565–
579, Aug. 2013, doi: 10.1016/j.imavis.2013.04.004.
[21] A. Kostusiak and P. Skrzypczyński, “On the
Efficiency of Population-Based Optimization in
Finding Best Parameters for RGB-D Visual
Odometry,” JAMRIS, vol. 13, no. 2, pp. 5–14, Jul.
2019, doi: 10.14313/JAMRIS/2-2019/13.
[22] A. Dib, “Vers un système de capture du mouvement
humain en 3D pour un robot mobile évoluant dans un
environnement encombré,” Doctorat en Intelligence
artificielle, Université de Lorraine, Nancy, France,
2016. [Online]. Available: https://hal.inria.fr/tel-
01333772
[23] Y. Ahmine, G. Caron, F. Chouireb, and E. M.
Mouaddib, “Continuous Scale-Space Direct Image
Alignment for Visual Odometry From RGB-D
Images,” IEEE Robot. Autom. Lett., vol. 6, no. 2, pp.
2264–2271, Apr. 2021, doi:
10.1109/LRA.2021.3061309.
[24] J. L. Blanco-Claraco, “A tutorial on SE (3)
transformation parameterizations and on-manifold
optimization,” Perception and Robotics Group,
University of Malaga, Spain, 2021, doi:
10.48550/ARXIV.2103.15980.
[25] S. Ahuja, “Lie Algebra to Lie Group Mapping,” 2015.
Accessed: Jan. 01, 2021. [Online]. Available:
https://math.stackexchange.com/questions/1312314/li
e-algebra-to-lie-group-mapping
[26] E.-G. Talbi, Metaheuristics, John Wiley&Sons., vol.
74. Hoboken, NJ, USA: John Wiley & Sons, Inc.,
2009. doi: 10.1002/9780470496916.
[27] E. H. Adelson, C. H. Anderson, J. R. Bergen, P. J. Burt,
and J. M. Ogden, “Pyramid methods in image
processing,” Radio Corporation of America Engineer
, Princeton, New Jersey, USA, vol. 29, no. 6, pp. 33–
41, 1984.
[28] E. N. Eriksen, “Monocular Visual Odometry for
Underwater Navigation,” TTK4900 – Master thesis,
Norwegian University of Science and Technology,
Trondheim, Norvège, 2020.
[29] J. Sturm, N. Engelhard, F. Endres, W. Burgard, and D.
Cremers, “A benchmark for the evaluation of RGB-D
SLAM systems,” in 2012 IEEE/RSJ International
Conference on Intelligent Robots and Systems,
Vilamoura-Algarve, Portugal, Oct. 2012, pp. 573–580.
doi: 10.1109/IROS.2012.6385773.
[30] W. Chen et al., “An Overview on Visual SLAM: From
Tradition to Semantic,” Remote Sensing, vol. 14, no.
13, p. 3010, Jun. 2022, doi: 10.3390/rs14133010.
[31] D. Prokhorov, D. Zhukov, O. Barinova, K. Anton, and
A. Vorontsova, “Measuring robustness of Visual
SLAM,” in 2019 16th International Conference on
Machine Vision Applications (MVA), Tokyo, Japan,
May 2019, pp. 1–6. doi:
10.23919/MVA.2019.8758020.
[32] Robert Roberts, Daniel Taylor, Juan Herrera, Juan
Castro, Mette Christensen. Integrating Virtual Reality
and Machine Learning in Education. Kuwait Journal
of Machine Learning, 2(1). Retrieved from
http://kuwaitjournals.com/index.php/kjml/article/view
/175
[33] Mr. Dharmesh Dhabliya, Dr.S.A.Sivakumar. (2019).
Analysis and Design of Universal Shift Register Using
Pulsed Latches . International Journal of New
Practices in Management and Engineering, 8(03), 10 -
16. https://doi.org/10.17762/ijnpme.v8i03.78
[34] Mandal, D., Shukla, A., Ghosh, A., Gupta, A., &
Dhabliya, D. (2022). Molecular dynamics simulation
for serial and parallel computation using leaf frog
algorithm. Paper presented at the PDGC 2022 - 2022
7th International Conference on Parallel, Distributed
and Grid Computing, 552-557.
doi:10.1109/PDGC56933.2022.10053161 Retrieved
from www.scopus.com