A stereo video segmentation algorithm combining disparity map and frame difference

In this paper, an algorithm to segment object from stereo video is presented. The algorithm combines the strong points of disparity map with frame difference. First, segment disparity maps to get the object segmentation template in different disparity lays. Then make frame difference extraction in the template regions to modify the object edge. After that the precise motion object has been gotten. The algorithm is low complexity and easy to realize. From the simulation results, we can conclude that the algorithm is efficient in segmentation when there is overlapping in different disparity motion objects or the background is moving.

[1]  Yao Wang,et al.  Video Processing and Communications , 2001 .

[2]  Antonio Ortega,et al.  Overlapped block disparity compensation with adaptive windows for stereo image coding , 2000, IEEE Trans. Circuits Syst. Video Technol..

[3]  Bertrand Chupeau,et al.  Depth-based segmentation , 1997, IEEE Trans. Circuits Syst. Video Technol..

[4]  Carlo Tomasi,et al.  Depth Discontinuities by Pixel-to-Pixel Stereo , 1999, International Journal of Computer Vision.

[5]  Ferran Marqués,et al.  Region-based representations of image and video: segmentation tools for multimedia services , 1999, IEEE Trans. Circuits Syst. Video Technol..

[6]  D. Scharstein,et al.  A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms , 2001, Proceedings IEEE Workshop on Stereo and Multi-Baseline Vision (SMBV 2001).