The Research And Application Of Mathematical Morphology Based On Soft Computing

The principles and properties of soft computing and mathematical morphology are reviewed. According to this, soft-multi-structural elements and Soft morphological operations based on soft computing are introduced. Based on the investigation of their properties and the links between the mentioned operators with classical morphological operators, some applications for image processing such as the morphological filters, edge detection are presented. Experimental results demonstrate that the ability of color image processing based on the morphological operations based on soft computing is superior to the classical one, and that the algorithm possesses excellent performance on the noise restrained.The work includes the following subjects:1) Based on the classical structural elements, a new structural element (Soft-multi-structural elements) is presented.A new filter is constructed based on it.Experimental results show that the new structural element is better than classical structural elements on the color image processing.2) Based on the T-S fuzzy system and soft-multi-structural elements mathematical morphology, new filter and edge detection are presented on color image processing of HSI.Our experimental results demonstrate that it is possible to improve the results in denoising by using the algorithm.3) In the space of HSV a new filter is presented on color morphology based on fuzzy combination and soft-multi-structural elements. Color image pixels are sorted by the value of fuzzy combination. Morphological filter based on the ranking of fuzzy combination can get better results than based on the ranking of HSV. Experimental results demonstrate that the ability of color processing based on the new morphology is superior to the classical one, and that the algorithm possesses excellent performance on the noise restrained.4) Based on Sugeno fuzzy integral and soft-multi-structural elements, new filter and edge detection are presented. Color image pixels are sorted by the value of Sugeno fuzzy integral. Comparing with other structural elements, soft-multi-structural elements are better on the noise restrained. Morphological transform based on the ranking of Sugeno fuzzy integral can get better results than based on the ranking of HSV. Experimental results demonstrate that the ability of color processing based on the new morphology is superior to the classical one, and that the algorithm possesses excellent performance on the noise restrained and obtained better results.