SlideShare a Scribd company logo
1
Feature Detection &
Extraction Using Parallel
Programming
04.09.2017
─
Arysha Channa
CT-056
BCIT
NEDUET
2
Table of Contents
Overview............................................................................................................3
Feature Detection ...........................................................................................3
Feature Extraction ........................................................................................3
Parallel Processing....................................................................................3
Goals..................................................................................................................4
Specifications.....................................................................................................4
Introduction.....................................................................................................4
Feature Detection (Classification).................................................................4
Feature Extraction (GP-GPU)....................................................................4
Parallel processing using OPENMP, SSE and CUDA............................................5
Configuration Using OPENMP.......................................................................5
Configuration Using SSE...........................................................................5
Configuration Using CUDA...................................................................6
Reference...........................................................................................................7
3
Overview
It has been estimated that the process of recognizing objects and
performing differentfeature detectionand extraction tasks requires
processorspeedto be 1 to 100 billion operations per second.
An image consists of differentfeatures. Features are basically, the
distinctive properties of an image or any input that help in differentiating the
objects from one another into differentcategories.
Feature detectionis the ability to break down an image into its components
i.e. its features, such as color, form, and motion. Moreover, if the detection
of various features occurs at once, or in parallel, it is known as parallel
processing.Thus, detectionof various features of an image simultaneously
is referred to as “Feature DetectionUsing Parallel Processing”.
Feature detectionmay include the information about the image to
comprehend what the objectis.
The major features are: Color, Form, and Motion
 Color detectionis recognized through cones in eye (red, green, and
blue)
 Form detectionuses parvo pathway [1] (responsible forfiguring out
the shape/outline of the object) by establishing the boundaries of the
object,the shape of the features.
 Motion Detectionutilizes magno pathway in the brain, the movement
of the object.
Parvo Pathway
- Vital for spatialresolution; efficientat identifying outboundaries of objects
- Good spatialresolution,but very poor temporalresolution
- Also,helps us to see in color
Magno Pathway
- Set of specialized cells that permitus to encrypt motion
- Very great temporal resolution;low spatialresolution
- Does not encode color;only motion
4
Feature Extraction [2] may include objectsegmentationusing features like
gray-level edges,texture edges,and shape attributes, depending on the
application’s requirements. It transforms arbitrary input data, such as
images, into a set of features. This helps in revealing unique parts of an
image as a compressedfeature vector.
Parallel Processing is the ability to processcolor,form, and motion of an
objectsimultaneously for feature detection.
Whereas,for feature extraction using different parallel processing
techniques and methods can accelerate computational problems.So,the
effectivenessof a multi-microprocessor [2] system where the tasks can be
distributed among differentprocessors will work in parallel.
Goals
1. To utilize parallel processing forfeature detectionand extraction
method to be executed at real-time with high-speed processing.
2. Identifyand extract distinctive features simultaneously.
3. Feature detectionand extraction are important componentof various
computervision applications.
Like, Image representation, image classification,and retrieval object
recognition and matching, 3Dscene reconstruction, motion tracking,
texture classification, robotlocalization and biometrics systems.
Specifications
I. INTRODUCTION
Feature detectionhas becomean influential tool in various fields of artificial
intelligence and is being applied immenselyin various other applications.
Thus, detecting image feature is an essential step for such applications.
One of the basic necessityfor feature detectionis to establish saline
feature correspondenceamong a collectionof differentimages sharing
similar features. This can be achieved by using classificationtask, where
feature details of test image are coordinated with all the trained image
5
features. The trained image features show a greater proportionof
correspondenceis consideredthe best match with the test images.[3, 4]
Another important area of research for image features is feature extraction.
Feature extraction is applied on CPU version, as the algorithms and the
codes are improvised and then executed using parallel processing resulting
in feature extraction in parallel. [4]
Then, using CUDA (based on the SIFT [4] language), a parallel
descriptoris constructed via SSE instructions, and the GPU version is also
implemented by parallel processing.
Using GPU-Parallel descriptorspeed up the process five times faster
than the CPU version. Whereas,CPU version accelerates by four and half
times than the SIFT.
As the feature-based approaches,like SIFT (Scale Invariant Feature
Transform) algorithms shows signs of robust performance;therefore,to
improve performance and make it faster, algorithms like PCA-SIFTand
SURF (Speed Up RobustFeatures) have beenintroduced. [4]
But, all the SIFT methods were hurdled by a higher-dimensional
descriptorwhich is extremely time-consuming and complexto function for
real-time processing.
In counter to all performance issues,parallel systems, [4] like:
OpenMP (an API- ApplicationProgram Interface),SSE (Streaming SIMD
Extension) and CUDA (Compute Unified Device Architecture) programming
are highly in demand for their stable high-speed algorithms to accomplish
fast and real-time feature extraction in objectfor images.
II. Parallel processing using OPENMP, SSE and CUDA
A. Parallelprocessing configurationusing OpenMP
OpenMP is used to make several threads for parallel programming
in a public, shared memoryenvironment. OpenMP comprises of
[4]: 1- compilerdirectives,2- a run-time library & 3- environmental
variables.
The drawback of this approach is that the OpenMP compiler
does not automatically assist analysis of all problems and
6
parallelization. Therefore,the pre-analysis of the sequential
program must directly handle the issues of data dependency,
synchronization initiated by parallelization.
B. Parallelprocessing configurationusing SSE
SSE are the extended versions of MMX instruction set, used in
Intel Pentium 4 and Core 2 Duo processorgenerations.SSE can
generate multiple SIMD (Single Instruction and Multiple Data) of
x87-bit floating point instructions via n-bit register(according to the
version used).
The SIMD instructions can be executed in an Intrinsic function, Automatic
Vectorization, or Inline Assembler.[4]
C. Parallelprocessing configurationusing CUDA
Graphics applications face the problems of the bottleneck.
Therefore, GPU was designed to settle this issue of bottlenecks.
GPU is a special-functionprocessing unit [4] with advanced
transistor densities and SIMD parallel hardware architecture. This
software and hardware architecture provide an independent
platform for CUDA by giving a programming environment to it, and
thus, reducing the need for considering graphics pipeline.
CUDA is designed using C language for general computing on
the GPU hardware and software.
GP-GPU (General Purpose Computation on GPUs) exhibits
real-time performance in applications with a large amount of data.
7
The figure shows the processof feature extraction using CUDA. For
parallel processing to be performed,each block produces alarge number
of threads to share data with other threads. [4]
References
[1] M. P.Ali Yoonessi,"Functional Assessmentof Magno,Parvoand Konio-CellularPathways;,"J
OphthalmicVisRes, vol.6, no.2, p. 119–126, 2011.
[2] T. N. M. Edward J.Delp,"Parallel processingforcomputervision," Robotvision, vol.336, 1982.
[3] A. A.A. H. A.A. M. Hassaballah,"FeaturesDetection," ImageFeaturesDetection,Description, pp.1-
36, 2016.
[4] Y. J. X. C.W. A. G. Junchul Kim,"Feature Extraction," A FastFeatureExtraction in ObjectRecognition,
p. 2, 2009.
.
8
Index
A
applications, 1
B
bottleneck, 1
C
classification, 1
Color, 1
CPU, 1
CUDA, 1
D
data, 1
detection, 1
E
extraction, 1
F
feature, 1
form, 1
functions, 1
G
GPU, 1
graphics, 1
I
Instructions, 1
M
magno, 1
Motion, 1
O
OPENMP, 1
P
Parallel, 1
parvo, 1
pathway, 1
process, 1
processing, 1
processor, 1
R
real-time, 1
S
SIFT, 1
SIMD, 1
SSE, 1
SURF, 1
T
techniques, 1
thread, 1

More Related Content

Similar to Feature detection & extraction

NVIDIA CUDA
NVIDIA CUDANVIDIA CUDA
NVIDIA CUDA
Jungsoo Nam
 
SMI_SNUG_paper_v10
SMI_SNUG_paper_v10SMI_SNUG_paper_v10
SMI_SNUG_paper_v10Igor Lesik
 
Choudhary2015
Choudhary2015Choudhary2015
Choudhary2015
Amit Kabra
 
Accelerating Real Time Applications on Heterogeneous Platforms
Accelerating Real Time Applications on Heterogeneous PlatformsAccelerating Real Time Applications on Heterogeneous Platforms
Accelerating Real Time Applications on Heterogeneous Platforms
IJMER
 
Deep Convolutional Neural Network acceleration on the Intel Xeon Phi
Deep Convolutional Neural Network acceleration on the Intel Xeon PhiDeep Convolutional Neural Network acceleration on the Intel Xeon Phi
Deep Convolutional Neural Network acceleration on the Intel Xeon Phi
Gaurav Raina
 
Deep Convolutional Network evaluation on the Intel Xeon Phi
Deep Convolutional Network evaluation on the Intel Xeon PhiDeep Convolutional Network evaluation on the Intel Xeon Phi
Deep Convolutional Network evaluation on the Intel Xeon Phi
Gaurav Raina
 
Thesis Report - Gaurav Raina MSc ES - v2
Thesis Report - Gaurav Raina MSc ES - v2Thesis Report - Gaurav Raina MSc ES - v2
Thesis Report - Gaurav Raina MSc ES - v2Gaurav Raina
 
Mod 2 hardware_graphics.pdf
Mod 2 hardware_graphics.pdfMod 2 hardware_graphics.pdf
Mod 2 hardware_graphics.pdf
DavidsonJebaseelan1
 
Distributed Database practicals
Distributed Database practicals Distributed Database practicals
Distributed Database practicals
Vrushali Lanjewar
 
Generating higher accuracy digital data products by model parameter
Generating higher accuracy digital data products by model parameterGenerating higher accuracy digital data products by model parameter
Generating higher accuracy digital data products by model parameterIAEME Publication
 
Parallel Processor for Graphics Acceleration
Parallel Processor for Graphics AccelerationParallel Processor for Graphics Acceleration
Parallel Processor for Graphics Acceleration
Sandip Jassar (sandipjassar@hotmail.com)
 
INOVA GIS Platform - TeleCAD-GIS & IGS (2018)
INOVA GIS Platform - TeleCAD-GIS & IGS (2018)INOVA GIS Platform - TeleCAD-GIS & IGS (2018)
INOVA GIS Platform - TeleCAD-GIS & IGS (2018)
Maksim Sestic
 
IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS An efficient-parallel-approach-fo...
IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS  An efficient-parallel-approach-fo...IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS  An efficient-parallel-approach-fo...
IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS An efficient-parallel-approach-fo...
IEEEBEBTECHSTUDENTPROJECTS
 
Real-Time Implementation and Performance Optimization of Local Derivative Pat...
Real-Time Implementation and Performance Optimization of Local Derivative Pat...Real-Time Implementation and Performance Optimization of Local Derivative Pat...
Real-Time Implementation and Performance Optimization of Local Derivative Pat...
IJECEIAES
 
High Performance Medical Reconstruction Using Stream Programming Paradigms
High Performance Medical Reconstruction Using Stream Programming ParadigmsHigh Performance Medical Reconstruction Using Stream Programming Paradigms
High Performance Medical Reconstruction Using Stream Programming Paradigms
QuEST Global (erstwhile NeST Software)
 
Distributed Traffic management framework
Distributed Traffic management frameworkDistributed Traffic management framework
Distributed Traffic management framework
Saurabh Nambiar
 
Spie2006 Paperpdf
Spie2006 PaperpdfSpie2006 Paperpdf
Spie2006 Paperpdf
Falascoj
 
Graphics processing unit ppt
Graphics processing unit pptGraphics processing unit ppt
Graphics processing unit ppt
Sandeep Singh
 
Orthogonal Matching Pursuit in 2D for Java with GPGPU Prospectives
Orthogonal Matching Pursuit in 2D for Java with GPGPU ProspectivesOrthogonal Matching Pursuit in 2D for Java with GPGPU Prospectives
Orthogonal Matching Pursuit in 2D for Java with GPGPU ProspectivesMatt Simons
 

Similar to Feature detection & extraction (20)

NVIDIA CUDA
NVIDIA CUDANVIDIA CUDA
NVIDIA CUDA
 
SMI_SNUG_paper_v10
SMI_SNUG_paper_v10SMI_SNUG_paper_v10
SMI_SNUG_paper_v10
 
Choudhary2015
Choudhary2015Choudhary2015
Choudhary2015
 
Accelerating Real Time Applications on Heterogeneous Platforms
Accelerating Real Time Applications on Heterogeneous PlatformsAccelerating Real Time Applications on Heterogeneous Platforms
Accelerating Real Time Applications on Heterogeneous Platforms
 
Deep Convolutional Neural Network acceleration on the Intel Xeon Phi
Deep Convolutional Neural Network acceleration on the Intel Xeon PhiDeep Convolutional Neural Network acceleration on the Intel Xeon Phi
Deep Convolutional Neural Network acceleration on the Intel Xeon Phi
 
Deep Convolutional Network evaluation on the Intel Xeon Phi
Deep Convolutional Network evaluation on the Intel Xeon PhiDeep Convolutional Network evaluation on the Intel Xeon Phi
Deep Convolutional Network evaluation on the Intel Xeon Phi
 
Thesis Report - Gaurav Raina MSc ES - v2
Thesis Report - Gaurav Raina MSc ES - v2Thesis Report - Gaurav Raina MSc ES - v2
Thesis Report - Gaurav Raina MSc ES - v2
 
Mod 2 hardware_graphics.pdf
Mod 2 hardware_graphics.pdfMod 2 hardware_graphics.pdf
Mod 2 hardware_graphics.pdf
 
Distributed Database practicals
Distributed Database practicals Distributed Database practicals
Distributed Database practicals
 
Generating higher accuracy digital data products by model parameter
Generating higher accuracy digital data products by model parameterGenerating higher accuracy digital data products by model parameter
Generating higher accuracy digital data products by model parameter
 
Parallel Processor for Graphics Acceleration
Parallel Processor for Graphics AccelerationParallel Processor for Graphics Acceleration
Parallel Processor for Graphics Acceleration
 
INOVA GIS Platform - TeleCAD-GIS & IGS (2018)
INOVA GIS Platform - TeleCAD-GIS & IGS (2018)INOVA GIS Platform - TeleCAD-GIS & IGS (2018)
INOVA GIS Platform - TeleCAD-GIS & IGS (2018)
 
IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS An efficient-parallel-approach-fo...
IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS  An efficient-parallel-approach-fo...IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS  An efficient-parallel-approach-fo...
IEEE 2014 MATLAB IMAGE PROCESSING PROJECTS An efficient-parallel-approach-fo...
 
Real-Time Implementation and Performance Optimization of Local Derivative Pat...
Real-Time Implementation and Performance Optimization of Local Derivative Pat...Real-Time Implementation and Performance Optimization of Local Derivative Pat...
Real-Time Implementation and Performance Optimization of Local Derivative Pat...
 
High Performance Medical Reconstruction Using Stream Programming Paradigms
High Performance Medical Reconstruction Using Stream Programming ParadigmsHigh Performance Medical Reconstruction Using Stream Programming Paradigms
High Performance Medical Reconstruction Using Stream Programming Paradigms
 
Distributed Traffic management framework
Distributed Traffic management frameworkDistributed Traffic management framework
Distributed Traffic management framework
 
Introduction to Blackfin BF532 DSP
Introduction to Blackfin BF532 DSPIntroduction to Blackfin BF532 DSP
Introduction to Blackfin BF532 DSP
 
Spie2006 Paperpdf
Spie2006 PaperpdfSpie2006 Paperpdf
Spie2006 Paperpdf
 
Graphics processing unit ppt
Graphics processing unit pptGraphics processing unit ppt
Graphics processing unit ppt
 
Orthogonal Matching Pursuit in 2D for Java with GPGPU Prospectives
Orthogonal Matching Pursuit in 2D for Java with GPGPU ProspectivesOrthogonal Matching Pursuit in 2D for Java with GPGPU Prospectives
Orthogonal Matching Pursuit in 2D for Java with GPGPU Prospectives
 

Recently uploaded

FIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdf
FIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdfFIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdf
FIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdf
FIDO Alliance
 
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
Neo4j
 
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...
Neo4j
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
sonjaschweigert1
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
mikeeftimakis1
 
Epistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI supportEpistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI support
Alan Dix
 
Secstrike : Reverse Engineering & Pwnable tools for CTF.pptx
Secstrike : Reverse Engineering & Pwnable tools for CTF.pptxSecstrike : Reverse Engineering & Pwnable tools for CTF.pptx
Secstrike : Reverse Engineering & Pwnable tools for CTF.pptx
nkrafacyberclub
 
UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4
DianaGray10
 
Pushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 daysPushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 days
Adtran
 
UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5
DianaGray10
 
PCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase TeamPCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase Team
ControlCase
 
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfObservability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Paige Cruz
 
Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?
Nexer Digital
 
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdfFIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance
 
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 previewState of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
Prayukth K V
 
FIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdfFIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance
 
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Albert Hoitingh
 
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
BookNet Canada
 
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
James Anderson
 
Essentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FMEEssentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FME
Safe Software
 

Recently uploaded (20)

FIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdf
FIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdfFIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdf
FIDO Alliance Osaka Seminar: Passkeys and the Road Ahead.pdf
 
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
 
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...
GraphSummit Singapore | The Future of Agility: Supercharging Digital Transfor...
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
 
Epistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI supportEpistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI support
 
Secstrike : Reverse Engineering & Pwnable tools for CTF.pptx
Secstrike : Reverse Engineering & Pwnable tools for CTF.pptxSecstrike : Reverse Engineering & Pwnable tools for CTF.pptx
Secstrike : Reverse Engineering & Pwnable tools for CTF.pptx
 
UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4
 
Pushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 daysPushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 days
 
UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5
 
PCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase TeamPCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase Team
 
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfObservability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
 
Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?
 
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdfFIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
 
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 previewState of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
 
FIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdfFIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdf
 
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
 
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
 
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
 
Essentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FMEEssentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FME
 

Feature detection & extraction

  • 1. 1 Feature Detection & Extraction Using Parallel Programming 04.09.2017 ─ Arysha Channa CT-056 BCIT NEDUET
  • 2. 2 Table of Contents Overview............................................................................................................3 Feature Detection ...........................................................................................3 Feature Extraction ........................................................................................3 Parallel Processing....................................................................................3 Goals..................................................................................................................4 Specifications.....................................................................................................4 Introduction.....................................................................................................4 Feature Detection (Classification).................................................................4 Feature Extraction (GP-GPU)....................................................................4 Parallel processing using OPENMP, SSE and CUDA............................................5 Configuration Using OPENMP.......................................................................5 Configuration Using SSE...........................................................................5 Configuration Using CUDA...................................................................6 Reference...........................................................................................................7
  • 3. 3 Overview It has been estimated that the process of recognizing objects and performing differentfeature detectionand extraction tasks requires processorspeedto be 1 to 100 billion operations per second. An image consists of differentfeatures. Features are basically, the distinctive properties of an image or any input that help in differentiating the objects from one another into differentcategories. Feature detectionis the ability to break down an image into its components i.e. its features, such as color, form, and motion. Moreover, if the detection of various features occurs at once, or in parallel, it is known as parallel processing.Thus, detectionof various features of an image simultaneously is referred to as “Feature DetectionUsing Parallel Processing”. Feature detectionmay include the information about the image to comprehend what the objectis. The major features are: Color, Form, and Motion  Color detectionis recognized through cones in eye (red, green, and blue)  Form detectionuses parvo pathway [1] (responsible forfiguring out the shape/outline of the object) by establishing the boundaries of the object,the shape of the features.  Motion Detectionutilizes magno pathway in the brain, the movement of the object. Parvo Pathway - Vital for spatialresolution; efficientat identifying outboundaries of objects - Good spatialresolution,but very poor temporalresolution - Also,helps us to see in color Magno Pathway - Set of specialized cells that permitus to encrypt motion - Very great temporal resolution;low spatialresolution - Does not encode color;only motion
  • 4. 4 Feature Extraction [2] may include objectsegmentationusing features like gray-level edges,texture edges,and shape attributes, depending on the application’s requirements. It transforms arbitrary input data, such as images, into a set of features. This helps in revealing unique parts of an image as a compressedfeature vector. Parallel Processing is the ability to processcolor,form, and motion of an objectsimultaneously for feature detection. Whereas,for feature extraction using different parallel processing techniques and methods can accelerate computational problems.So,the effectivenessof a multi-microprocessor [2] system where the tasks can be distributed among differentprocessors will work in parallel. Goals 1. To utilize parallel processing forfeature detectionand extraction method to be executed at real-time with high-speed processing. 2. Identifyand extract distinctive features simultaneously. 3. Feature detectionand extraction are important componentof various computervision applications. Like, Image representation, image classification,and retrieval object recognition and matching, 3Dscene reconstruction, motion tracking, texture classification, robotlocalization and biometrics systems. Specifications I. INTRODUCTION Feature detectionhas becomean influential tool in various fields of artificial intelligence and is being applied immenselyin various other applications. Thus, detecting image feature is an essential step for such applications. One of the basic necessityfor feature detectionis to establish saline feature correspondenceamong a collectionof differentimages sharing similar features. This can be achieved by using classificationtask, where feature details of test image are coordinated with all the trained image
  • 5. 5 features. The trained image features show a greater proportionof correspondenceis consideredthe best match with the test images.[3, 4] Another important area of research for image features is feature extraction. Feature extraction is applied on CPU version, as the algorithms and the codes are improvised and then executed using parallel processing resulting in feature extraction in parallel. [4] Then, using CUDA (based on the SIFT [4] language), a parallel descriptoris constructed via SSE instructions, and the GPU version is also implemented by parallel processing. Using GPU-Parallel descriptorspeed up the process five times faster than the CPU version. Whereas,CPU version accelerates by four and half times than the SIFT. As the feature-based approaches,like SIFT (Scale Invariant Feature Transform) algorithms shows signs of robust performance;therefore,to improve performance and make it faster, algorithms like PCA-SIFTand SURF (Speed Up RobustFeatures) have beenintroduced. [4] But, all the SIFT methods were hurdled by a higher-dimensional descriptorwhich is extremely time-consuming and complexto function for real-time processing. In counter to all performance issues,parallel systems, [4] like: OpenMP (an API- ApplicationProgram Interface),SSE (Streaming SIMD Extension) and CUDA (Compute Unified Device Architecture) programming are highly in demand for their stable high-speed algorithms to accomplish fast and real-time feature extraction in objectfor images. II. Parallel processing using OPENMP, SSE and CUDA A. Parallelprocessing configurationusing OpenMP OpenMP is used to make several threads for parallel programming in a public, shared memoryenvironment. OpenMP comprises of [4]: 1- compilerdirectives,2- a run-time library & 3- environmental variables. The drawback of this approach is that the OpenMP compiler does not automatically assist analysis of all problems and
  • 6. 6 parallelization. Therefore,the pre-analysis of the sequential program must directly handle the issues of data dependency, synchronization initiated by parallelization. B. Parallelprocessing configurationusing SSE SSE are the extended versions of MMX instruction set, used in Intel Pentium 4 and Core 2 Duo processorgenerations.SSE can generate multiple SIMD (Single Instruction and Multiple Data) of x87-bit floating point instructions via n-bit register(according to the version used). The SIMD instructions can be executed in an Intrinsic function, Automatic Vectorization, or Inline Assembler.[4] C. Parallelprocessing configurationusing CUDA Graphics applications face the problems of the bottleneck. Therefore, GPU was designed to settle this issue of bottlenecks. GPU is a special-functionprocessing unit [4] with advanced transistor densities and SIMD parallel hardware architecture. This software and hardware architecture provide an independent platform for CUDA by giving a programming environment to it, and thus, reducing the need for considering graphics pipeline. CUDA is designed using C language for general computing on the GPU hardware and software. GP-GPU (General Purpose Computation on GPUs) exhibits real-time performance in applications with a large amount of data.
  • 7. 7 The figure shows the processof feature extraction using CUDA. For parallel processing to be performed,each block produces alarge number of threads to share data with other threads. [4] References [1] M. P.Ali Yoonessi,"Functional Assessmentof Magno,Parvoand Konio-CellularPathways;,"J OphthalmicVisRes, vol.6, no.2, p. 119–126, 2011. [2] T. N. M. Edward J.Delp,"Parallel processingforcomputervision," Robotvision, vol.336, 1982. [3] A. A.A. H. A.A. M. Hassaballah,"FeaturesDetection," ImageFeaturesDetection,Description, pp.1- 36, 2016. [4] Y. J. X. C.W. A. G. Junchul Kim,"Feature Extraction," A FastFeatureExtraction in ObjectRecognition, p. 2, 2009. .
  • 8. 8 Index A applications, 1 B bottleneck, 1 C classification, 1 Color, 1 CPU, 1 CUDA, 1 D data, 1 detection, 1 E extraction, 1 F feature, 1 form, 1 functions, 1 G GPU, 1 graphics, 1 I Instructions, 1 M magno, 1 Motion, 1 O OPENMP, 1 P Parallel, 1 parvo, 1 pathway, 1 process, 1 processing, 1 processor, 1 R real-time, 1 S SIFT, 1 SIMD, 1 SSE, 1 SURF, 1 T techniques, 1 thread, 1