Abstract: The paper presents a method of image edge detection based on multi-structure and multi-grade weighted combination of mathematical morphology，in which the idea of morphological filter is used and on the anti-noise operators of morphological based on single-structure are improved.In this method，the fixed mean with the peak-value SNR（Signal-to-Noise） is replaced，which is tested a self-adaptive method.The experiment result compared with traditional morphological methods shows that this method remains good detected edge of image and can improve the denoising power as well.

摘要： 基于改进的单结构元素抗噪型形态边缘检测算子和形态滤波思想，提出了一种多结构元素的多路加权合成形态边缘检测算法。在该算法中，采用峰值信噪比代替固定均值来确定加权参数，是一种自适应方法。与传统的边缘检测算法的对比实验表明，该算法图像边缘检测效果较好，降噪能力也得到了提高。