A detail-preserving filtering module based on mathematical morphology

Morphological filtering is a new and developing nonlinear filtering technique. The filtering principle of traditional morphological filters based on fitting the appointed structuring element to each objects of an image, which lead to damage image details seriously. So, this paper proposed a CS filtering model based on multiple structuring elements. Firstly use a contour-structuring element to extract the noise areas, and then use a connected structuring element to eliminate all the pixels in the noise areas at one time. This model distinguishes noise and valid objects by measuring objects´ morphological extension. Many tests have been done to compare the filtering effect between the CS filtering model, the traditional morphological filters, Wiener filter and median filter. Experimental results indicate that the CS filtering model is really a detail-preserving filter.