Image segmentation with 2-D maximum entropy based on comprehensive learning particle swarm optimization

A novel optimal thresholding algorithm for image segmentation based on 2-D histogram is presented in this paper. The 2-D maximum entropy method not only makes use of the distribution of the gray information, but also takes advantage of the spatial neighbor information with using the 2-D histogram of the image. It can get ideal segmentation results from the images with lower signal noise ratio (SNR). However, the time-consuming computation is often an obstacle for this method to be used in real time application systems. By analyzing the theory of entropy threshold segmentation, the comprehensive learning particle swarm optimization (CLPSO) is then used to counteract premature convergence, so that we can obtain the maximum entropy. CLPSO algorithm is used successfully to solve the problem of 2-D maximum entropy. The experimental results of images segmentation are illustrated to show that the proposed method can get ideal segmentation result with less computation demand.