For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Robust Moving Object Extraction and Tracking Method Based on Matching Position Constraints
Tetsuya OKUDA Yoichi TOMIOKA Hitoshi KITAZAWA
IEICE TRANSACTIONS on Information and Systems
Publication Date: 2015/08/01
Online ISSN: 1745-1361
Type of Manuscript: PAPER
Category: Image Recognition, Computer Vision
moving object extraction, tracking, shape information, motion segmentation,
Full Text: PDF>>
Object extraction and tracking in a video image is basic technology for many applications, such as video surveillance and robot vision. Many moving object extraction and tracking methods have been proposed. However, they fail when the scenes include illumination change or light reflection. For tracking the moving object robustly, we should consider not only the RGB values of input images but also the shape information of the objects. If the objects' shapes do not change suddenly, matching positions on the cost matrix of exclusive block matching are located nearly on a line. We propose a method for obtaining the correspondence of feature points by imposing a matching position constraint induced by the shape constancy. We demonstrate experimentally that the proposed method achieves robust tracking in various environments.