A study of color image segmentation based on stochastic expectation maximization algorithm in HSV model

This paper addresses a study of target segmentation on two color images based on SEM algorithm and region growing algorithm. Background image and target-existing image are converted from RGB space to HSV space. The Euclid distance between these two transformed images in HSV space is computed and compared with that in RGB space. To segment the target region from the background, SEM algorithm is applied. Then the MAP criterion is used for further segmentation. With certain prior knowledge about the size of the target, final segmentation result is got by region growing algorithm. The result of simulation shows that these segmentation methods are very efficient when used together