Abstract: Preserving image details while filtering noise has been a hot research topic. The effect of morphological filtering depends on the selection of shapes and sizes of structural elements. The CB morphology to some extent weakens the impact of the structural elements on the results. But the outline of the large structures leads to noise amplification, and the smaller structure contour results in noise filtering ineffectively. It presents an improved CB morphological filter, that uses small structural elements, CB morphology and mathematical morphological. The results show that the effect of the filter to filter out salt and pepper noise is better than the median filtering or morphological noise filter, and that the filter can reserve better image details at the same time.

Key words: mathematical morphology, Contour Bougie（CB） morphology, image filtering, structural element

摘要： 如何在滤除噪声的同时保护图像的细节一直是一个研究热点。基于数学形态学的滤波效果依赖于结构元素形状和尺寸的选取，基于轮廓结构元素形态学（简称CB形态学）在一定程度上弱化了结构元素对处理结果的影响，但是大结构轮廓会导致噪声放大，小结构轮廓噪声滤除效果不佳。提出一种改进CB形态学滤波器，利用小结构元素、CB形态学、数学形态学运算实现对图像椒盐噪声的滤除。实验结果表明该滤波器滤除椒盐噪声的效果优于中值滤波或一般形态学滤噪，且能够在滤除噪声的同时较好地保留图像细节。

关键词: 数学形态学, 轮廓结构元素形态学（CB形态学）, 图像滤波, 结构元素