Efficient Particle Filtering for a Non-rigid Object Based on PCA about Changes of Its Shape and Motion

Ikuhisa MITSUGAMI  Koh KAKUSHO  Michihiko MINOH  

D - Abstracts of IEICE TRANSACTIONS on Information and Systems (Japanese Edition)   Vol.J92-D   No.8   pp.1270-1278
Publication Date: 2009/08/01
Online ISSN: 1881-0225
Print ISSN: 1880-4535
Type of Manuscript: Special Section PAPER (Special Section on Image Recognition and Understanding)
object tracking,  particle filter,  PCA,  non-rigid objects,  

Full Text(in Japanese): PDF(449.8KB)
>>Buy this Article

This paper proposes a novel tracking method for non-rigid objects that is based on the particle filtering. The particle filtering method prepares a parametric model of the object's motion and shape, and estimates the parameters in each frame. In this estimation, although these parameters may changes related to one another especially in the case of the non-rigid objects, they are calculated independently so that the calculation is not efficient. Considering this inefficiency, the proposed method focuses on the relation between the motion and shape of the non-rigid object, obtains the relation preliminarily by offline tracking using enough number of particles, and then generates the particles efficiently based on the relation at the online step. By this method, the tracking works well in spite of the reduction of the particles. Experimental results of goldfish tracking show the effectiveness.