SlideShare a Scribd company logo
1 of 24
IMPLEMENTING CAMSHIFT ON A
MOBILE ROBOT FOR
PERSON TRACKING AND PURSUIT
Somar Boubou, Asuki Kouno,
Einoshin Suzuki
Kyushu University, Japan
@LIMER 2011, Vancouver
2011, Dec 11th
1
OUTLINE
 Objective and applications
 Problem
 Tracking approaches
 Mean-shift and Camshift
 Related work
 Robot hardware and software
 Experiments
 Results
 Discussion
2
OBJECTIVE
A Camshift implementation on mobile robotic
system for tracking and pursuing a moving person
with a camera.
Although researches are very promising, the
problem is still far away from being solved. 3
 Museum guidance
[Yamazaki, K- 2009] (CHI
2009)
 Hospital assistance
[Panasonic Medication Dispensing and
Delivery Robots] (2010)
 Socially assistive robot.
[M. Montemerlo 2002]
APPLICATIONS
ENVIRONME
NT
ROBOT
University student
office.
Hand made robot equipped with one
camera
• Visual detection and tracking
component.
Detect, track and pursuit.
PROBLEM
Task
5
ASSUMPTIO
NS
CONSTRAINTS
DUE TO ON-
BOARD
IMPLEMENTATIO
N
 Tracked person wears two wrist
bands on his both foots.
 Person must be 2m far or less from
the robot, so it can clearly detect
him.
The system has :
• to be robust regarding external
noise (e.g., illumination,
background changes and low-
resolution camera input).
• a sufficient processing speed to
PROBLEM
6
 Model-based tracking: Where the geometric structure
of a human body can be represented as 2D or 3D model.
 Region-based tracking: Considers the human body
as a combination of some blobs, representing various body
parts.
 Active-contour-based tracking: Aims at directly
extracting the shape of the subjects by a representation of
the bounding contour of the object and keeps dynamically
updating it.
 Feature-based tracking: Abandons the idea of
tracking objects as a whole, this tracking approach uses
TRACKING
APPROCHES
[Liang Wang 2002]
7
MEAN-SHIFT
 Introduced by Y. Cheng. in “Mean Shift, Mode
Seeking, and Clustering” PAMI 1995.
 Mean-shift algorithm
climbs the gradient of a
probability distribution to
find the nearest peak.
8
MEAN-SHIFT
Mean shift
vector (follows
density
gradient)
Reference data set
Query point
Center of Mass
x
h
xx
k
h
xx
kx
xm
n
i
i
i
n
i
i








 







 





1
2
2
1
'
'
)(
Mean shift vector with
kernel k
[Dorin Comaniciu 2000]
9
MEAN-SHIFT
Reference data set
Query point
Center of Mass
x
h
xx
k
h
xx
kx
xm
n
i
i
i
n
i
i








 







 





1
2
2
1
'
'
)(
Mean shift vector with
kernel k
[Dorin Comaniciu 2000]
Convergence to
cluster center
1
CONTINUOUSLY ADAPTIVE MEAN-
SHIFT (CAMSHIFT) [GR Bradski. In1998]
Camshift is used for video sequence,
 CAMSHIFT operates on a probability distribution
image derived from color histograms.
 CAMSHIFT calculates the centroid of the 2D color
probability distribution within its 2D window of
calculation.
 Re-centers the window, then calculates the area for
the next window size.
It differs from Mean-shift: Search window adjusts its
size dynamically.
11
FLOW CHART OF PROPPOSED
METHOD
Camshift-based tracking
method
Frame Hue Backprojecti
on
ROIs Tracking
Threshol
dHistogra
m
cf. HSV-based tracking use only threshold to define
1
RELATED WORK
 [Fhed 2009]: People Tracking via a Modified Camshift
Algorithm
Fixed platform with static background
 [Simon 2009]: Most Salient Region Tracking
Manual initialization by selecting a rectangle around the
target of interest
 [Feyrer 1999]: Detection, Tracking, and Pursuit of
Humans with an Autonomous Mobile Robot
Used HSI color model combined with motion and contour
information for person detection.
 [Saravanakumar 2011]: Multiple Object Tracking Using
HSV Color Spaces
1
OUR WORK
 System Initialization and Detection
Human detection aims at segmenting regions
corresponding to people from the rest of the image.
 Tracking and pursuit
Tracking over time typically involves matching objects in
consecutive frames using features such as points, lines or
blobs.
Originality: Implementing CAMSHIFT for real robot.
1
①SYSTEM INITIALIZATION AND
DETECTION
 Converts the sequence of images into
backprojection images.
 First robot loads the saved
wearable mark template, analyses
it then creates the target
histogram.
 Use mean-shift to define ROIs (Region Of Interest)
we used Camshift to detect the wearable marks
coordinates (u, v).
1
α
Go Forward
Turn Right
Turn
Left
𝟓
𝟖
× 𝑼 𝒄𝒐𝒈 − 𝑼 𝒄𝒆𝒏𝒕𝒆𝒓
millisec
Stabilization
range
Action
range
We use the ROI’s windows
defined during detection step to
initialize Camshift instants
used for tracking the wearable
marks.
②TRACKING/ PURSUIT
W : Image width.
FOV : Field of View. FOV
𝑈 𝑎 = α ×
𝑊
𝐹𝑂𝑉
1
ROBOT
HARDWARE 6 Touch
sensors
8 IR sensors
USB camera
USB camera
IR sensors
Touch sensors
PandaBoard
Arduino LED unit
Motors
Input MPUs Output
PandaBoard
• 1GHz CPU
• 1GB Memory
• 16 GB SDHC
card
Arduino
• 16 MHz
• 8 KB memory
The two independent components exchange information via a fast inter
process serial communication. 1
ROBOT SOFTWARE AND
CONTROLLER PROGRAM
DIAGRAM
The diagram to the right
explains the controller
initialization and
Initialize camera
input
If object
template
new
Load object
template
Find Object/s
Select object
Save object template
Track
object/s
Start
End
Main controlling algorithms, for both
PandaBoard and Arduino board,
are written using C language.
We also implemented
Camshift algorithm,
which is publicly
available from openCV
library.
Ye
s
No
1
 Wearable marks (i.e., Wrist bands) with size 6 × 8cm.
 Variables manual calibration of camshift parameters:
Hue (27 < H < 32).
saturation (100 < S < 255).
Value (200 < V < 255).
 10 experiments for each method
where robot pursuit a person
walking inside an office in
approximate path of 11.25 meter
length with average speed of
14cm/sec.
EXPERIMENTAL CONDITIONS
1
DESIGN OF THE EXPERIMENTS
We compared Camshift with a HSV color based
tracking methods by evaluating the performance of
the robot in two levels:
 Visual tracking.
 Robot pursuit performance.
Also we compared execution
time for both methods:
• Camshift (363 milisecond)
• HSV (297 millisecond)
Two frames continuous miss-track
EXPEREMINTAL RESULTS
 As figure shows Camshift is more robust and stable
since the variance of its results is smaller.
 Temporal distribution of tracking failures affects
pursuit performance
0%
20%
40%
60%
80%
100%
Visual tracking Pursuit
Successrate
Camshift
HSV-based tracking
2
DISSCUSSION
 Although, Camshift method is computationally
slower, it outperformed the HSV color based
tracking.
 Camshift method is much more robust against
different illumination conditions.
 Both methods show better tracking results with
larger targets (e.g., tracking the T-shirt or trousers).
2
 During the experiments we realized that the HSV
color based method had serious difficulties in the
first part of our path.
PERFORMANCE VEDIO - 2x speed
2
Thank you for listening

More Related Content

What's hot

Detection and Tracking of Moving Object: A Survey
Detection and Tracking of Moving Object: A SurveyDetection and Tracking of Moving Object: A Survey
Detection and Tracking of Moving Object: A SurveyIJERA Editor
 
Object detection technique using bounding box algorithm for
Object detection technique using bounding box algorithm forObject detection technique using bounding box algorithm for
Object detection technique using bounding box algorithm forVESIT,Chembur,Mumbai
 
Multiple Object Tracking
Multiple Object TrackingMultiple Object Tracking
Multiple Object TrackingRainakSharma
 
Fahad Fazal Elahi Guraya
Fahad Fazal Elahi GurayaFahad Fazal Elahi Guraya
Fahad Fazal Elahi Gurayakimberleychen
 
Object tracking survey
Object tracking surveyObject tracking survey
Object tracking surveyRich Nguyen
 
motion and feature based person tracking in survillance videos
motion and feature based person tracking in survillance videosmotion and feature based person tracking in survillance videos
motion and feature based person tracking in survillance videosshiva kumar cheruku
 
Visual Object Tracking: review
Visual Object Tracking: reviewVisual Object Tracking: review
Visual Object Tracking: reviewDmytro Mishkin
 
multiple object tracking using particle filter
multiple object tracking using particle filtermultiple object tracking using particle filter
multiple object tracking using particle filterSRIKANTH DANDE
 
Object Detection & Tracking
Object Detection & TrackingObject Detection & Tracking
Object Detection & TrackingAkshay Gujarathi
 
TRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCES
TRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCESTRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCES
TRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCESPraveen Pallav
 
Object tracking
Object trackingObject tracking
Object trackingchirase44
 
Object tracking a survey
Object tracking a surveyObject tracking a survey
Object tracking a surveyHaseeb Hassan
 
Moving object detection
Moving object detectionMoving object detection
Moving object detectionManav Mittal
 
Moving object detection using background subtraction algorithm using simulink
Moving object detection using background subtraction algorithm using simulinkMoving object detection using background subtraction algorithm using simulink
Moving object detection using background subtraction algorithm using simulinkeSAT Publishing House
 
Moving object detection in video surveillance
Moving object detection in video surveillanceMoving object detection in video surveillance
Moving object detection in video surveillanceAshfaqul Haque John
 
Video Surveillance Systems For Traffic Monitoring
Video Surveillance Systems For Traffic MonitoringVideo Surveillance Systems For Traffic Monitoring
Video Surveillance Systems For Traffic MonitoringMeridian Media
 

What's hot (20)

Detection and Tracking of Moving Object: A Survey
Detection and Tracking of Moving Object: A SurveyDetection and Tracking of Moving Object: A Survey
Detection and Tracking of Moving Object: A Survey
 
Object detection technique using bounding box algorithm for
Object detection technique using bounding box algorithm forObject detection technique using bounding box algorithm for
Object detection technique using bounding box algorithm for
 
Multiple Object Tracking
Multiple Object TrackingMultiple Object Tracking
Multiple Object Tracking
 
Object tracking
Object trackingObject tracking
Object tracking
 
Fahad Fazal Elahi Guraya
Fahad Fazal Elahi GurayaFahad Fazal Elahi Guraya
Fahad Fazal Elahi Guraya
 
Object tracking survey
Object tracking surveyObject tracking survey
Object tracking survey
 
motion and feature based person tracking in survillance videos
motion and feature based person tracking in survillance videosmotion and feature based person tracking in survillance videos
motion and feature based person tracking in survillance videos
 
Camshift
CamshiftCamshift
Camshift
 
Visual Object Tracking: review
Visual Object Tracking: reviewVisual Object Tracking: review
Visual Object Tracking: review
 
multiple object tracking using particle filter
multiple object tracking using particle filtermultiple object tracking using particle filter
multiple object tracking using particle filter
 
Object Detection & Tracking
Object Detection & TrackingObject Detection & Tracking
Object Detection & Tracking
 
TRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCES
TRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCESTRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCES
TRACKING OF PARTIALLY OCCLUDED OBJECTS IN VIDEO SEQUENCES
 
Object tracking
Object trackingObject tracking
Object tracking
 
Object tracking a survey
Object tracking a surveyObject tracking a survey
Object tracking a survey
 
Moving object detection1
Moving object detection1Moving object detection1
Moving object detection1
 
Moving object detection
Moving object detectionMoving object detection
Moving object detection
 
Background subtraction
Background subtractionBackground subtraction
Background subtraction
 
Moving object detection using background subtraction algorithm using simulink
Moving object detection using background subtraction algorithm using simulinkMoving object detection using background subtraction algorithm using simulink
Moving object detection using background subtraction algorithm using simulink
 
Moving object detection in video surveillance
Moving object detection in video surveillanceMoving object detection in video surveillance
Moving object detection in video surveillance
 
Video Surveillance Systems For Traffic Monitoring
Video Surveillance Systems For Traffic MonitoringVideo Surveillance Systems For Traffic Monitoring
Video Surveillance Systems For Traffic Monitoring
 

Similar to Implementing Camshift on a Mobile Robot for Person Tracking and Pursuit_ICDM

Simultaneous Mapping and Navigation For Rendezvous in Space Applications
Simultaneous Mapping and Navigation For Rendezvous in Space ApplicationsSimultaneous Mapping and Navigation For Rendezvous in Space Applications
Simultaneous Mapping and Navigation For Rendezvous in Space ApplicationsNandakishor Jahagirdar
 
SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS
 SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS  SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS
SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS Nandakishor Jahagirdar
 
Modelling of walking humanoid robot with capability of floor detection and dy...
Modelling of walking humanoid robot with capability of floor detection and dy...Modelling of walking humanoid robot with capability of floor detection and dy...
Modelling of walking humanoid robot with capability of floor detection and dy...ijfcstjournal
 
Development of wearable object detection system &amp; blind stick for visuall...
Development of wearable object detection system &amp; blind stick for visuall...Development of wearable object detection system &amp; blind stick for visuall...
Development of wearable object detection system &amp; blind stick for visuall...Arkadev Kundu
 
Interactive full body motion capture using infrared sensor network
Interactive full body motion capture using infrared sensor networkInteractive full body motion capture using infrared sensor network
Interactive full body motion capture using infrared sensor networkijcga
 
LANE DETECTION USING IMAGE PROCESSING IN PYTHON
LANE DETECTION USING IMAGE PROCESSING IN PYTHONLANE DETECTION USING IMAGE PROCESSING IN PYTHON
LANE DETECTION USING IMAGE PROCESSING IN PYTHONIRJET Journal
 
Design and implementation of video tracking system based on camera field of view
Design and implementation of video tracking system based on camera field of viewDesign and implementation of video tracking system based on camera field of view
Design and implementation of video tracking system based on camera field of viewsipij
 
Interactive Full-Body Motion Capture Using Infrared Sensor Network
Interactive Full-Body Motion Capture Using Infrared Sensor Network  Interactive Full-Body Motion Capture Using Infrared Sensor Network
Interactive Full-Body Motion Capture Using Infrared Sensor Network ijcga
 
Landmines Detection by Robots presentation
Landmines Detection by Robots  presentationLandmines Detection by Robots  presentation
Landmines Detection by Robots presentationAhmed Abdelaziz
 
Camouflage Color Changing Robot For Military Purpose
Camouflage Color Changing Robot For Military PurposeCamouflage Color Changing Robot For Military Purpose
Camouflage Color Changing Robot For Military PurposeHitesh Shinde
 
Camouflage color changing robot for miltary purpose
Camouflage color changing robot for miltary purposeCamouflage color changing robot for miltary purpose
Camouflage color changing robot for miltary purposeAtharvaPathak13
 
A Robot Collision Avoidance Method Using Kinect and Global Vision
A Robot Collision Avoidance Method Using Kinect and Global VisionA Robot Collision Avoidance Method Using Kinect and Global Vision
A Robot Collision Avoidance Method Using Kinect and Global VisionTELKOMNIKA JOURNAL
 
ROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTING
ROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTINGROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTING
ROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTINGIAEME Publication
 
Real time-image-processing-applied-to-traffic-queue-detection-algorithm
Real time-image-processing-applied-to-traffic-queue-detection-algorithmReal time-image-processing-applied-to-traffic-queue-detection-algorithm
Real time-image-processing-applied-to-traffic-queue-detection-algorithmajayrampelli
 
Follow Me Robot Technology
Follow Me Robot TechnologyFollow Me Robot Technology
Follow Me Robot Technologyijsrd.com
 
Home automation using MATLAB image processing
Home automation using MATLAB image processingHome automation using MATLAB image processing
Home automation using MATLAB image processingNarayanlalMenariya
 

Similar to Implementing Camshift on a Mobile Robot for Person Tracking and Pursuit_ICDM (20)

Simultaneous Mapping and Navigation For Rendezvous in Space Applications
Simultaneous Mapping and Navigation For Rendezvous in Space ApplicationsSimultaneous Mapping and Navigation For Rendezvous in Space Applications
Simultaneous Mapping and Navigation For Rendezvous in Space Applications
 
SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS
 SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS  SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS
SIMULTANEOUS MAPPING AND NAVIGATION FOR RENDEZVOUS IN SPACE APPLICATIONS
 
Modelling of walking humanoid robot with capability of floor detection and dy...
Modelling of walking humanoid robot with capability of floor detection and dy...Modelling of walking humanoid robot with capability of floor detection and dy...
Modelling of walking humanoid robot with capability of floor detection and dy...
 
H028038042
H028038042H028038042
H028038042
 
Development of wearable object detection system &amp; blind stick for visuall...
Development of wearable object detection system &amp; blind stick for visuall...Development of wearable object detection system &amp; blind stick for visuall...
Development of wearable object detection system &amp; blind stick for visuall...
 
Interactive full body motion capture using infrared sensor network
Interactive full body motion capture using infrared sensor networkInteractive full body motion capture using infrared sensor network
Interactive full body motion capture using infrared sensor network
 
LANE DETECTION USING IMAGE PROCESSING IN PYTHON
LANE DETECTION USING IMAGE PROCESSING IN PYTHONLANE DETECTION USING IMAGE PROCESSING IN PYTHON
LANE DETECTION USING IMAGE PROCESSING IN PYTHON
 
Design and implementation of video tracking system based on camera field of view
Design and implementation of video tracking system based on camera field of viewDesign and implementation of video tracking system based on camera field of view
Design and implementation of video tracking system based on camera field of view
 
Luigy Bertaglia Bortolo - Poster Final
Luigy Bertaglia Bortolo - Poster FinalLuigy Bertaglia Bortolo - Poster Final
Luigy Bertaglia Bortolo - Poster Final
 
Interactive Full-Body Motion Capture Using Infrared Sensor Network
Interactive Full-Body Motion Capture Using Infrared Sensor Network  Interactive Full-Body Motion Capture Using Infrared Sensor Network
Interactive Full-Body Motion Capture Using Infrared Sensor Network
 
Landmines Detection by Robots presentation
Landmines Detection by Robots  presentationLandmines Detection by Robots  presentation
Landmines Detection by Robots presentation
 
Camouflage Color Changing Robot For Military Purpose
Camouflage Color Changing Robot For Military PurposeCamouflage Color Changing Robot For Military Purpose
Camouflage Color Changing Robot For Military Purpose
 
Camouflage color changing robot for miltary purpose
Camouflage color changing robot for miltary purposeCamouflage color changing robot for miltary purpose
Camouflage color changing robot for miltary purpose
 
A Robot Collision Avoidance Method Using Kinect and Global Vision
A Robot Collision Avoidance Method Using Kinect and Global VisionA Robot Collision Avoidance Method Using Kinect and Global Vision
A Robot Collision Avoidance Method Using Kinect and Global Vision
 
ROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTING
ROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTINGROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTING
ROBOTOR AN AUTONOMOUS VEHICLE FOR TARGET DETECTION AND SHOOTING
 
Real time-image-processing-applied-to-traffic-queue-detection-algorithm
Real time-image-processing-applied-to-traffic-queue-detection-algorithmReal time-image-processing-applied-to-traffic-queue-detection-algorithm
Real time-image-processing-applied-to-traffic-queue-detection-algorithm
 
30120140506012 2
30120140506012 230120140506012 2
30120140506012 2
 
30120140506012 2
30120140506012 230120140506012 2
30120140506012 2
 
Follow Me Robot Technology
Follow Me Robot TechnologyFollow Me Robot Technology
Follow Me Robot Technology
 
Home automation using MATLAB image processing
Home automation using MATLAB image processingHome automation using MATLAB image processing
Home automation using MATLAB image processing
 

Recently uploaded

Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...Christo Ananth
 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escortsranjana rawat
 
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...ranjana rawat
 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCall Girls in Nagpur High Profile
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Dr.Costas Sachpazis
 
Porous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingPorous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingrakeshbaidya232001
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...Soham Mondal
 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Serviceranjana rawat
 
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICSHARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICSRajkumarAkumalla
 
UNIT-V FMM.HYDRAULIC TURBINE - Construction and working
UNIT-V FMM.HYDRAULIC TURBINE - Construction and workingUNIT-V FMM.HYDRAULIC TURBINE - Construction and working
UNIT-V FMM.HYDRAULIC TURBINE - Construction and workingrknatarajan
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSKurinjimalarL3
 
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINEMANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINESIVASHANKAR N
 
Introduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxIntroduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxupamatechverse
 
Processing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptxProcessing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptxpranjaldaimarysona
 
SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )Tsuyoshi Horigome
 
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)Suman Mia
 
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptxthe ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptxhumanexperienceaaa
 

Recently uploaded (20)

Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
 
Call Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCR
Call Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCRCall Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCR
Call Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCR
 
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
 
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINEDJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
 
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
 
Porous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingPorous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writing
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
 
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICSHARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
 
UNIT-V FMM.HYDRAULIC TURBINE - Construction and working
UNIT-V FMM.HYDRAULIC TURBINE - Construction and workingUNIT-V FMM.HYDRAULIC TURBINE - Construction and working
UNIT-V FMM.HYDRAULIC TURBINE - Construction and working
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
 
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINEMANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
 
Introduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxIntroduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptx
 
Processing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptxProcessing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptx
 
SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )
 
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
 
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptxthe ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
 

Implementing Camshift on a Mobile Robot for Person Tracking and Pursuit_ICDM

  • 1. IMPLEMENTING CAMSHIFT ON A MOBILE ROBOT FOR PERSON TRACKING AND PURSUIT Somar Boubou, Asuki Kouno, Einoshin Suzuki Kyushu University, Japan @LIMER 2011, Vancouver 2011, Dec 11th 1
  • 2. OUTLINE  Objective and applications  Problem  Tracking approaches  Mean-shift and Camshift  Related work  Robot hardware and software  Experiments  Results  Discussion 2
  • 3. OBJECTIVE A Camshift implementation on mobile robotic system for tracking and pursuing a moving person with a camera. Although researches are very promising, the problem is still far away from being solved. 3
  • 4.  Museum guidance [Yamazaki, K- 2009] (CHI 2009)  Hospital assistance [Panasonic Medication Dispensing and Delivery Robots] (2010)  Socially assistive robot. [M. Montemerlo 2002] APPLICATIONS
  • 5. ENVIRONME NT ROBOT University student office. Hand made robot equipped with one camera • Visual detection and tracking component. Detect, track and pursuit. PROBLEM Task 5
  • 6. ASSUMPTIO NS CONSTRAINTS DUE TO ON- BOARD IMPLEMENTATIO N  Tracked person wears two wrist bands on his both foots.  Person must be 2m far or less from the robot, so it can clearly detect him. The system has : • to be robust regarding external noise (e.g., illumination, background changes and low- resolution camera input). • a sufficient processing speed to PROBLEM 6
  • 7.  Model-based tracking: Where the geometric structure of a human body can be represented as 2D or 3D model.  Region-based tracking: Considers the human body as a combination of some blobs, representing various body parts.  Active-contour-based tracking: Aims at directly extracting the shape of the subjects by a representation of the bounding contour of the object and keeps dynamically updating it.  Feature-based tracking: Abandons the idea of tracking objects as a whole, this tracking approach uses TRACKING APPROCHES [Liang Wang 2002] 7
  • 8. MEAN-SHIFT  Introduced by Y. Cheng. in “Mean Shift, Mode Seeking, and Clustering” PAMI 1995.  Mean-shift algorithm climbs the gradient of a probability distribution to find the nearest peak. 8
  • 9. MEAN-SHIFT Mean shift vector (follows density gradient) Reference data set Query point Center of Mass x h xx k h xx kx xm n i i i n i i                         1 2 2 1 ' ' )( Mean shift vector with kernel k [Dorin Comaniciu 2000] 9
  • 10. MEAN-SHIFT Reference data set Query point Center of Mass x h xx k h xx kx xm n i i i n i i                         1 2 2 1 ' ' )( Mean shift vector with kernel k [Dorin Comaniciu 2000] Convergence to cluster center 1
  • 11. CONTINUOUSLY ADAPTIVE MEAN- SHIFT (CAMSHIFT) [GR Bradski. In1998] Camshift is used for video sequence,  CAMSHIFT operates on a probability distribution image derived from color histograms.  CAMSHIFT calculates the centroid of the 2D color probability distribution within its 2D window of calculation.  Re-centers the window, then calculates the area for the next window size. It differs from Mean-shift: Search window adjusts its size dynamically. 11
  • 12. FLOW CHART OF PROPPOSED METHOD Camshift-based tracking method Frame Hue Backprojecti on ROIs Tracking Threshol dHistogra m cf. HSV-based tracking use only threshold to define 1
  • 13. RELATED WORK  [Fhed 2009]: People Tracking via a Modified Camshift Algorithm Fixed platform with static background  [Simon 2009]: Most Salient Region Tracking Manual initialization by selecting a rectangle around the target of interest  [Feyrer 1999]: Detection, Tracking, and Pursuit of Humans with an Autonomous Mobile Robot Used HSI color model combined with motion and contour information for person detection.  [Saravanakumar 2011]: Multiple Object Tracking Using HSV Color Spaces 1
  • 14. OUR WORK  System Initialization and Detection Human detection aims at segmenting regions corresponding to people from the rest of the image.  Tracking and pursuit Tracking over time typically involves matching objects in consecutive frames using features such as points, lines or blobs. Originality: Implementing CAMSHIFT for real robot. 1
  • 15. ①SYSTEM INITIALIZATION AND DETECTION  Converts the sequence of images into backprojection images.  First robot loads the saved wearable mark template, analyses it then creates the target histogram.  Use mean-shift to define ROIs (Region Of Interest) we used Camshift to detect the wearable marks coordinates (u, v). 1
  • 16. α Go Forward Turn Right Turn Left 𝟓 𝟖 × 𝑼 𝒄𝒐𝒈 − 𝑼 𝒄𝒆𝒏𝒕𝒆𝒓 millisec Stabilization range Action range We use the ROI’s windows defined during detection step to initialize Camshift instants used for tracking the wearable marks. ②TRACKING/ PURSUIT W : Image width. FOV : Field of View. FOV 𝑈 𝑎 = α × 𝑊 𝐹𝑂𝑉 1
  • 17. ROBOT HARDWARE 6 Touch sensors 8 IR sensors USB camera USB camera IR sensors Touch sensors PandaBoard Arduino LED unit Motors Input MPUs Output PandaBoard • 1GHz CPU • 1GB Memory • 16 GB SDHC card Arduino • 16 MHz • 8 KB memory The two independent components exchange information via a fast inter process serial communication. 1
  • 18. ROBOT SOFTWARE AND CONTROLLER PROGRAM DIAGRAM The diagram to the right explains the controller initialization and Initialize camera input If object template new Load object template Find Object/s Select object Save object template Track object/s Start End Main controlling algorithms, for both PandaBoard and Arduino board, are written using C language. We also implemented Camshift algorithm, which is publicly available from openCV library. Ye s No 1
  • 19.  Wearable marks (i.e., Wrist bands) with size 6 × 8cm.  Variables manual calibration of camshift parameters: Hue (27 < H < 32). saturation (100 < S < 255). Value (200 < V < 255).  10 experiments for each method where robot pursuit a person walking inside an office in approximate path of 11.25 meter length with average speed of 14cm/sec. EXPERIMENTAL CONDITIONS 1
  • 20. DESIGN OF THE EXPERIMENTS We compared Camshift with a HSV color based tracking methods by evaluating the performance of the robot in two levels:  Visual tracking.  Robot pursuit performance. Also we compared execution time for both methods: • Camshift (363 milisecond) • HSV (297 millisecond) Two frames continuous miss-track
  • 21. EXPEREMINTAL RESULTS  As figure shows Camshift is more robust and stable since the variance of its results is smaller.  Temporal distribution of tracking failures affects pursuit performance 0% 20% 40% 60% 80% 100% Visual tracking Pursuit Successrate Camshift HSV-based tracking 2
  • 22. DISSCUSSION  Although, Camshift method is computationally slower, it outperformed the HSV color based tracking.  Camshift method is much more robust against different illumination conditions.  Both methods show better tracking results with larger targets (e.g., tracking the T-shirt or trousers). 2  During the experiments we realized that the HSV color based method had serious difficulties in the first part of our path.
  • 23. PERFORMANCE VEDIO - 2x speed 2
  • 24. Thank you for listening

Editor's Notes

  1. Good afternoon every one, My name is Somar Boubou from Kyushu university Japan and Today I’ll present my work titled “Implementing Camshift on a Mobile Robot for Person Tracking and Pursuit” Co-authors are Asuki Kouno and my supervisor Einoshin Suzuki.
  2. At first let’s see the outlines of this presentation. I’ll start by Introduction and applications, Problem statement, followed by an overview of tracking algorithms and related work, Then I’ll explain the hardware and the software of the robot with our implemented method. At the end I’ll present the experiments, the results and the discussion.
  3. A Camshift implementation on mobile robotic system for tracking and pursuing a moving person with a monocular camera. Although researches are very promising, the problem is still far away from being solved.
  4. museum guidance, hospital assistance, or socially assistive robot.
  5. -Model-based tracking. Where the geometric structure of a human body can be represented as stick figure, 2-D contour or volumetric model. So body segments can be approximated as lines, 2-D ribbons and 3-D volumes accordingly. -Region-based tracking. Considers the human body as a combination of some blobs, representing various body parts such as head, torso and four limbs. -Active-contour-based tracking. Aims at directly extracting the shape of the subjects. The idea is to have a representation of the bounding contour of the object and keep dynamically updating it over time. -Feature-based tracking Abandons the idea of tracking objects as a whole, this tracking approach uses features such as distinguishable points or lines on the object to realize the tracking task.
  6. In mean-shift you have what’s known as reference data set. It’s the data set which you want to cluster (mean shift is a clustering algorithm) Another data set known as query data set. This correspond to the data set of coordinates that you would like to assign to cluster center. The way that mean shift is works as follows. Suppose that the red plus sign is your query data point. Then you draw a circle around it. Then compute center of mass in that circle shown by green plus sign. Then compute What is known the mean shift vector which basically follows the gradient of the density function. Then we shift the query point to the green plus sign. If we look to the formula of meanshift vector it looks a lot like kernel density estimation formula. K is divertive from kernel function
  7. Repeat until converge to class center
  8. When tracking a colored object, CAMSHIFT operates on a color probability distribution image derived from color histograms. CAMSHIFT calculates the centroid of the 2D color probability distribution within its 2D window of calculation, re-centers the window, then calculates the area for the next window size. Thus, we needn’t calculate the color probability distribution over the whole image, but can instead restrict the calculation of the distribution to a smaller image region surrounding the current CAMSHIFT window. This tends to result in large computational savings when flesh color does not dominate the image. We refer to this feedback of calculation region size as the Coupled CAMSHIFT algorithm.
  9. Histogram intersection is scale variant
  10. In vision-based tracking field, Camshift algorithm has already been used for face-tracking. It is, however, not reported to be successfully used for a person tracking on-board. Although modified Camshift was implemented by Fhed et. al. [6] for a fixed platform with static background to track people indoor, we are the first to implement Camshift for mobile robots. Simon et. al. [7] suggested a new MSR tracker method for finding salient regions of tracked objects. They claim that suggested method is applicable to track different kinds of objects without adjusting parameters. However it needs a manual initialization by selecting a rectangle around the target of interest. They also claimed that the system provides all conditions which are necessary to use it on mobile robots. However experiments were never done on physical robots. [Feyrer 1999] presents a system which is able to visually detect human faces, to track them by controlling a robot head, and to pursue a detected person by means of driving movements. They used HSI combined with motion and contour information for person detection. An example of implementing HSV on a fixed platform for multiple object tracking is introduced by [Saravanakumar 2011].
  11. The upper half of Figure shows that visual field of the robot is divided into three ranges. Each range, the rightmost and leftmost range is defined as action range of angle . The remaining range in the middle is called a stabilization range. The frame in Figure 2 is divided into three areas corresponding to the three ranges represented in the visual field of the robot. Those three areas of the frame are defined by widths Ua and Us. where W is set to be 360 pixel, is set to 40 and FOV have a mean 112 and variance 2. Simple rules were applied for the pursuit processing. Firstly, COG (Center of Gravity) of tracked multi-targets is calculated. Secondly, Robot keeps moving forward to pursuit detected human while COG stays in stabilization range and turns right or left if COG moved to one of the action ranges. Turning duration depends on the horizontal distance between COG and the frame center
  12. This decoupled architecture allows us to achieve a fast image processing even during quick movements of the person, while the second component is able to independently control output units and read sensors data in an appropriate processing frequency. The size of the robot is 20 22 10 centimeter
  13. Ubuntu 10.10 is installed on the PandaBoard. Main controlling algorithms, for both PandaBoard and Arduino board, are written using C language. We also implemented Camshift algorithm, which is publicly available from openCV library, for tracking. Arduino board program is responsible for controlling motors and receiving sensor signals while PandaBoard program is responsible for image processing. Figure shows a diagram of the controlling program on PandaBoard.
  14. It’s clearly shows that Camshift is much more appropriate to be implemented for robotics tracking and pursuit application compared to the HSV color based method.
  15. (Since the first part is near the window, we think that sunlight reflected on the metal surfaces could be responsible about that, but more detailed experiments are needed to prove it.)