中文摘要：

针对标记分水岭对生活彩色服装图片分割效果不佳的缺点，提出了基于分水岭算法的最大相似性区域合并算法。首先对彩色生活图片进行去噪处理，然后以彩色梯度图代替传统的灰度梯度图进行分水岭粗分割，最后结合人工标记区域合并算法对彩色服装图像进行了后期处理。结果表明：W—MSRM算法对于彩色生活图片的分割效果优于传统分水岭算法。结论：该算法能有效地实现目标对象和背景对象分离。

英文摘要：

Considering that the marker watershed for living color clothing image has poor segmen- tation results, this paper proposed Watershed Maximal Similarity algorithm based on Region Mer- ging. Firstly, to reduce the living color image noise, and then perform watershed coarse segmentation to color gradient map instead of traditional gray gradient map. Finally, conduct a post processing of region merging of artificial marker for Living Color Image to separate an object and a background. The experimental results show that： the effect of W_MSRM algorithm for living color image is better than the traditional watershed approach.