在介绍适用于二值图象的软化形态学的定义及特性的基础上，将软化形态学概念加以推广，使之适用于灰度图象，并提出了一种基软化形态学的边缘检测方法－SMD，还把此方法与其他的基于匀边形态学方法进行了比较，实验结果表明该方法具有很强的抑制噪声能力。

Mathematical morphology is an efficacious theory and method, which used in image processing and pattern recognition, especially in image edge detection. This paper introduces the definition of soft morphology, then presents a robust edge detector based on it SMD (Soft Morphological Detector),and compared with other edge detectors based on crisp morphology. The results of experiments show that this new detector SMD has very strong robustness, and performs better than other morphology detectors.