Fuzzy Mathematical Morphology And Its Application On Image Processing

As we all know, structure information is one of the most important characteristics during the processing of an image. It is easy to see that if we get the structure information, the processing time will be much less. Due to its special ability on structural image, mathematical morphology has more advatage over other image processing methods. This dissertation mainly discusses the properties and its applications of fuzzy mathematical morphology. The major contributions of this dissertation are summarized as follows.(l)Analyze abilities of fuzzy geodesic morphological operators over their a -cuts. Morphological operating P truly means transforming an image A into a new image P(A, B) via a little one B. Let A and B are both fuzzy sets, P(A, B) is also a fuzzy set. Theoretically, fuzzy mathematical morphological operators fulfill the decomposition theorem. Though the fuzzy operators are smoothly transformed from binary ones , it is necessary to discuss their abilities over their α -cuts because of their characteristics . The decomposition and reconstruction problems of fuzzy geodesic morphological operators, proposed by I.Bloch, are investigated. The relations between fuzzy geodesic morphological operators and their α -cuts are discussed and accordingly the important conclusion is revealed that fuzzy geodesic morphological operators can be reconstructed or decomposed using their a -cuts with appropriate t-norms and t-conorms. (2) Propose a new algorithm for white blood cell detection based on fuzzy' mathematical morphology. The algorithm use HLS color space instead of RGB ones and associate it with fuzzy mathematical morphology. Experimental results here demonstrate the very superiority of this new algorithm. Meanwhile, during some automatic cell recognition system, we should count the number of some kind of cells, especially white blood cells. Hence, a modified Fisher discriminant is proposed at first. Then maximum scatter difference classifier (MSDs) which based on the new discriminant is derived. It is showed that when parameter C in the MSDs is approaching infinity, a new kind of classifier called large margin linear projection classifier...