Image Thresholding Based on 2-Dimensional Gray Entropy and Chaotic Particle Swarm Algorithm

To consider simultaneously the histogram probability information and the uniformity of within-cluster gray level in the 2-dimensional maximum entropy thresholding method, the 2-dimensional Shannon gray entropy and Tsallis gray entropy thresholding methods are proposed based on gray level-gradient histogram in this article. First, the Shannon gray entropy and Tsallis gray entropy were defined and the one-dimensional thresholding methods were given. Then 2-dimensional Shannon gray entropy and Tsallis gray entropy thresholding formulae and their fast recursive algorithms were derived, and the chaotic particle swarm optimization algorithm was used to search the best thresholds. Lots of experiments were done and the results show that, compared with the thresholding method based on improved 2-dimensional maximum entropy and particle swarm optimization, the obtained segmented images using suggested method can reflect the edge, texture and details of the original images with more accuracy.