2007 IEEE International Conference on Image Processing - San Antonio, Texas, U.S.A. - September 16-19, 2007

Technical Program

Paper Detail

Paper:TA-P4.12
Session:Video Object Segmentation and Tracking I
Time:Tuesday, September 18, 09:50 - 12:30
Presentation: Poster
Title: A NEW STEREO CORRESPONDENCE METHOD FOR SNAKE-BASED OBJECT SEGMENTATION
Authors: Ashraf Alattar; PaiChai University 
 Jong Whan Jang; PaiChai University 
Abstract: In this paper, we propose a new method for generating excellent external energy for snake-based object segmentation methods in stereo images. Our method first generates an edge-based disparity map by performing stereo correspondence between multi-level edge maps of the stereo image pair. Only edges of similar strength are considered for matching. To filter the disparity map for edges of the object of interest, the method estimates the object’s disparity value by matching the pattern of edges of the region of interest in the left image against candidate patterns in the right image. The filtered edge map is then used to generate external energy for the snake. The proposed method has been tested on two snake models and results show a noticeable enhancement on performance of the snake when compared with other methods.



©2016 Conference Management Services, Inc. -||- email: webmaster@icip2007.com -||- Last updated Friday, August 17, 2012