Study In Color Noise Image Edge Detection Algorithm Based On The Mathematical Morphology

Edge is an important feature of the image, including the important informations. Theedge which extracted by the edge detection technology, has laid an important foundationfor the subsequent image segmentation and recognition. But along with the progress of thetimes, the development of digital technology, the images are almost color images.Andcolor images contains rich informations than the gray image, So the edge detection studiesof color images are more practical significance.And the mathematical morphology theory,as a kind of image processing and analysis tools has been widely used in shaperecognition,feature extraction,image denoising,edge detection,image segmentation, imageprocessing and so on. With the further research of image edge detection algorithm fromthe point of view of mathematical morphology theory,then puting the morphology ideasand nonlinear methods apply for color noise image edge detection. The research works inthis paper is as following.Frist,analyzed and discussed the basic gray-scale morphological edge detectionalgorithmic,then put forward two improved edge detection methods for the situation ofnoise and noiseless.Experimental results show that the two methods effect of edgeextraction has been greatly improved.Second,based on the research of gray image edge detection,then going deep into thecolor image,using double channel denoising way and the multi-structure and multi-scalemodel to effectively extract the different types geometry edge for H S and I.This methodcan get better denoising performance and accurate position, and the edge is complete,clear,smooth.Then,proposed a method of color image edge detection based on mathematicalmorphology and NSCT (Nonsubsampled Contourlet Transform) to detect the edge ofcolor noising image,after processing the edge through the method of AEO edgeoptimization,then fusing them through NSCT inverse transformation,Finally using EBO toprocess the edge. The experiments show that the proposed method can obtain more edgedetails.