A novel color texture-based segmentation algorithm is proposed. Many powerful color segmentation algorithms such as JSEG (J-SEGmentation) suffer from over segmentation.  An improved JSEG method called improved contrast JSEG (IC-JSEG) is developed to construct the contrast map to obtain the basic contours of the homogeneous regions in the image. A two serial type-based filtering and a noise-protected edge detector are adopted to remove the noise and enhance the edge strength to provide a better contrast map. Based on the combination of improved contrast map and the original J map in JSEG, seed growing-merging method is used to segment the image. Experiments on both natural color-texture images and color medical images show promising results.

 Project Sponsors:

 Collaborators:

 Graduate Students:

 Yuchou Chang

1. Y. Chang, D.J. Lee, and Y. Wang, “Color-Texture Segmentation of Medical Images Based on Local Contrast Information”, IEEE Symposium on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB 2007), Honolulu, Hawaii, USA, April 1-5, 2007.