Image Segmentation Using Correlative Histogram Modeled by Gaussian Mixture

In this paper we address the problem of gray image segmentation. Our approach falls in category of histogram based thresholding methods. From image we first construct a correlative histogram, based on intensity of image pixels and the average intensity of pixel neighbourhood. The proposed histogram is more informative than common intensity histogram for segmentation. Then we model the obtained histogram using a mixture of Gaussian functions. We estimate the parameters for Gaussian mixtures using particle swarm optimization algorithm. The result of segmentation confirms that the proposed method outperforms existing thresholding methods.