Regarding Head Shape

Author : Shun-Cho Jacky Yuk
Publisher : Open Dissertation Press
ISBN 13 : 9781361434369
Total Pages : pages
Book Rating : 4.4/5 (343 download)

DOWNLOAD NOW!

Download or read book A Probabilistic Framework for Real-Time Head Shape Detection and Tracking written by Shun-Cho Jacky Yuk and published by Open Dissertation Press. This book was released on 2017-01-27 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: This dissertation, "A Probabilistic Framework for Real-time Head Shape Detection and Tracking" by Shun-cho, Jacky, Yuk, 郁順祖, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Abstract of thesis entitled AProbabilisticFrameworkforReal-TimeHeadShapeDetection andTracking submitted by Jacky, Shun-ChoYUK for the Degree of Master of Philosophy at the University of Hong Kong in August 2007 Vision based real-time human tracking involves the automatic detection and track- ing of human bodies in video sequences. This task is not trivial due to the difficulties in identifying articulate human bodies in crowded and cluttered environments. Most of the existing approaches are either too computationally intensive or unable to han- dle crowded situations. This dissertation presents a robust human head detection and tracking system based on a probabilistic head shape detection framework with decen- tralized tracking approach. A fast and robust probabilistic shape contour matching framework is firstly devel- opedtodetectandlocateheadshapeobjectsinaveryefficientmanner. Theframework firstly applies a fast hough voting, and a pruning is performed according to the hough votes in order to eliminate most of the non-head locations. The remaining potential head candidates are then verified by a novel probabilistic shape template matching method. The fast hough pruning together with the efficient shape matching approach ensure that the framework is able to perform in real-time. The detected head objects are then tracked by decentralized trackers. Here, a de- centralizedtrackerreferstothetrackerthattracksexactlyoneobject. Essentially, each newly detected object will instantiate an individual tracker, which tracks the object and destroys itself when the object disappears. Two trackers communicate with each other only when they are getting close enough. This approach simplifies the compe- tition of targets between trackers, and is more efficient than the centralized approach whosetimecomplexitydependsheavilyonthenumberoftrackedobjects. Inthisway, each tracker can focus on its own target head to improve tracking robustness, and the efficiency of the tracking is reserved for real-time application. The system has been tested with several challenging digital surveillance video se- quences. Experimental results show that this approach is robust and very efficient for multiple head tracking in real-time even under crowded and clutter environments. (310 words) DOI: 10.5353/th_b3955886 Subjects: Computer vision