A Rapid Anomalous Region Extraction Method by Iterative Projection onto Kernel Eigen Space

Satoshi KAWABATA  Sinsaku HIURA  Kosuke SATO  

Publication
D - Abstracts of IEICE TRANSACTIONS on Information and Systems (Japanese Edition)   Vol.J91-D   No.11   pp.2673-2683
Publication Date: 2008/11/01
Online ISSN: 1881-0225
DOI: 
Print ISSN: 1880-4535
Type of Manuscript: PAPER
Category: 
Keyword: 
background subtraction,  dynamic background,  eigen space,  robust projection,  

Full Text(in Japanese): PDF(1.1MB)
>>Buy this Article


Summary: 
In computer vision, background subtraction method is widely used to extract a changing region in a scene. The method determines the region by comparing an input image with the background image obtained in advance. However, it is difficult to simply apply this method to a scene with moving background object, because such object may be extracted as changing region. Therefore, a method have been proposed to estimate both current background image and occluding object region simultaneously by using iterative projection to eigenspace. On the other hand, the image complement method using eigenspace have been extended to non-linear subspace by using the kernel trick, but however, such existing method takes a large computational cost. Therefore, In this paper, we propose a method for rapid simultaneous estimation of a background image and occluded region in non-linear space, using the kernel trick and iterative projection.