Abstract: An improved algorithm of multi-scale morphological edge detection is proposed.Edges of different size are detected by using different scale operator.And edge images are combined with the way of evidence syncretic fusion.Experiment on pathology image，the more ideal image edges are obtained under the environment of existing noise by interating the edges of various sizes.This paper compares the algorithm with others，the results demonstrate the feasibility and effectiveness，the algorithm can get distinct edges and noise is removed efficiently.

摘要： 提出了一种改进的多尺度形态边缘检测算法。用不同尺度的结构元素分别检测出图像的不同尺寸的边缘信息，然后采用证据加权的融合方法对不同尺寸的边缘图像进行融合。通过对病理显微图像的实验，在噪声存在的条件下得到较为理想的图像边缘。与其他边缘检测算法进行比较，结果表明该算法在有效地消除噪声的同时，能够取得较好的边缘检测效果。