Non-supervised image segmentation based on multiobjective optimization

The segmentation process based on the optimization of one criterion only does not work well for a lot of images. In many cases, even when equipped with the optimal value of the threshold of its single criterion, the segmentation program does not produce a satisfactory result. In this paper, we propose to use the multiobjective optimization approach to find the optimal thresholds of two criteria: the within-class criterion and the overall probability of error criterion. In addition we develop a new variant of Simulated Annealing adapted to continuous problems to solve the histogram Gaussian curve fitting problem. Six examples of test images are presented to compare the efficiency of our segmentation method, called Combination of Segmentation Objectives (CSO), based on the multiobjective optimization approach, with that of two classical competing methods: Otsu method and Gaussian curve fitting method. From the viewpoints of visualization, object size and image contrast, our experimental results show that the segmentation method based on multiobjective optimization performs better than the Otsu method and the method based on Gaussian curve fitting.