关键词: 图像分割, 子区域相似度, 轮廓逼近, 区域生长

Abstract: Introducing the traditional region growing method into a contour approximating method, a new segmentation algorithm based on subregion similarity was presented. Firstly, an initial polygon contour was brought in by human-computer interactive process. Then the region of interest was obtained by using a sub-region similarity which served as driving force of the contour evolution. The experimental results show that the algorithm is of more anti-noise ability than the region growing method, and can divide up the region of interest efficiently.

Key words: image segmentation, sub-region similarity, contour approximating, region growing