Adaptive Spatially Neighborhood Information Gaussian Mixture Model for Image Segmentation

One of the important characteristics of an image is that neighborhood pixels are highly correlated. In other words, these neighboring pixels possess similar feature values and the probability that they belong to the same cluster is great. Unfortunately,the application of Gaussian mixture model to image segmentation has not been taken into account spatial information except for intensity values, which could lead to misclassification on the boundaries and inhomogeneous regions with noise. In order to solve this problem, a new image segmentation method using adaptive spatially neighborhood information Gaussian mixture model without any control paremeters is proposed in this paper. Firstly, an adaptive spatial information function is defined to deal with the neighbour pixel of spatial correlatation,which is not only effective to deal with noise, but also to reserve well edge property. Secondly, it designs the neighbour information weighted class probabilities of every pixel according to Bayesian rules and proves that these class probabilities satisfy two norms of polarity and spatial continuity. Finally, an expectation maximization algorithm is used to obtain iterative formula of E-step and M-step as an optimization method. The experiments by synthetic images and real images demonstrate that the proposed method can obtain a better classification result and less effect on the noise.