The Research And Application Of Real Number Morphology

Mathematical morphology, which plays an important role in image processing, isa methodology of non-linear. It is a science based on strict mathematical theories. Ithas been widely used in the area of image processing such as noise suppression, edgedetection, image segmentation, feature extraction and so on.This article mainly researches improvement methods of mathematicalmorphology and its application to image filtering and image edge detection. Based onthe analysis of traditional mathematical morphology’s problem, this article proposesreal number structure elements and theories of real number morphology based on realnumber structure elements, and use the theories to solve problems of traditionalmorphology.Base on real number morphology, this article researches application of the realmorphology in image filtering, and use differential evolution algorithm to search thebest real morphological structure element. In this article, we research the fitnessfunction and propose Ration of Filter Function. It is used to measure the effect ofimage filtering and is used as fitness function of evolutionary algorithm. For problemsof differential evolution algorithm, this article proposes a differential evolutionalgorithm which can reflects the group learning the process of evolution, and appliesit to real morphological filtering.As the traditional methods such as Sobel, Robert, Prewitt, Log, Canny and so onare sensitive to noise. And the algorithm of edge detection based on traditionalmathematical morphology can not use spatial information effectively, so they aredifficult to detect complex edge feature. This article proposes multi_perspectivesalgorithm of edge detection based on real number structural element, this algorithmcan use distance and multiple perspectives completely. At last the experimental resultsshow that this algorithm can filter noise successfully and is local efficient for complexedge detection in the processing of images which have complex edge. This article putsresearch into application in the field of medical image processing, the experimentsdemonstrate the real number morphology of the feasibility and practicality.