Motion Based Image Segmentation with Unsupervised Bayesian Learning

An algorithm using Bayesian on-line learning for object based video image segmentation is proposed in this paper. First the strengths of image pixel's spatial location, color and motion segments are fused in one framework for image clustering and segmentation. Here the appropriate modeling of Probability Distribution Functions(PDF) of each feature cluster is obtained through Gaussian Distribution. In this paper unsupervised Bayesian learning is implemented to identify these distribution parameters. The online Bayesian learning process is carried out with the previous clustered image pixels information and feature clusters Gaussian PDF information. This algorithm has shown good results on different video files.

[1]  Narendra Ahuja,et al.  Hand gesture recognition and face detection in images , 2000 .

[2]  Paul Smith,et al.  Layered motion segmentation and depth ordering by tracking edges , 2004, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  A. Murat Tekalp,et al.  Region-Based Parametric Motion Segmentation Using Color Information , 1998, Graph. Model. Image Process..

[4]  Mubarak Shah,et al.  Object based segmentation of video using color, motion and spatial information , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.

[5]  Ishwar K. Sethi,et al.  Region Growing Using Online Learning , 2003, CISST.

[6]  Azriel Rosenfeld,et al.  Accurate dense optical flow estimation using adaptive structure tensors and a parametric model , 2003, IEEE Trans. Image Process..

[7]  Fernand Meyer,et al.  A novel approach to depth ordering in monocular image sequences , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).