Research Of Color Image Edge Detection Algorithm Based On Mathematical Morphology

This thesis studies color image edge detection techniques and gives some newcolor image edge detection models with the advantages of mathematical morphologyand various color space.Chapter1briefly introduces the background, significance and development historyof the color image edge detection, and the development history of the mathematicalmorphology; presents the problems and development trend of the edge detection.Chapter2introduces several basic color spaces, the transformation between themand the main methods of color image edge detection.Chapter3introduces the basic theory of mathematical morphology, and points outthe problems for using mathematical morphology in color image edge detection andhow to resolve them.Chapter4gives an improved algorithm for color image edge detection based onfuzzy morphological morphology. This algorithm extends the fuzzy morphology fromthe scalar operation for grayscale image to vector operation for color image, and thefuzzy enhancement formula is improved consequently. Moreover, the amoebastructural elements are introduced for edge extraction.In chapter5, for solving the problems of the vector ordering in RGB space, edgelocation and denoising of the color image edge detection, a color image edge detectionmodel based on the morphological amoebas (adaptive structuring elements) and RGBspace and HSV space is presented. Firstly, the amoeba structural elements arecalculated in the HSV space which overcomes the shortcomings of unreasonableselection of traditional morphological structure elements. Secondly, the vector orderingin RGB space is given according to the metric in HSV space. Thirdly, the edgeintensity is defined in RGB space by calculating the minimum distance between pixelsin the amoeba structural elements obtained, which not only avoids the inaccurate edgelocation in HSV space, but also improves the noise suppression. Finally, the idea of theCanny operator is used to get single pixel edge.Chapter6gives a conclusion of the thesis and presents some thinking andexpectation about the color image edge detection.