Fast automatic image segmentation based on Bayesian decision-making theory

As the precondition of image recognition, the effective image segmentation plays the significant role of the following image processing. In this paper, it is proposed to apply Bayesian decision-making theory based on minimum error probability to gray image segmentation. On the assumption that the gray values accord with the probability distribution of Gaussian finite mixture model in image feature space, EM algorithm is used to estimate the parameters of mixture model. In order to improve the convergence speed of EM algorithm, a novel and feasible method called weighted equal interval sampling is presented to obtain the contracted sample set. Consequently, the computation task of EM algorithm is greatly reduced and efficiency is improved. An approximate MMIC algorithm of Bayesian Information Criterion is employed to determine quickly how many regions should be segmented on a given gray image. The automatic image segmentation can be executed with the method mentioned above. We demonstrate the effectiveness and feasibility of our method on a set of natural and synthetic images.