Edge Detection Algorithm Based On Mathematical Morphology For Medical Noisy Image

Medical image plays an important role in the detection and clinical diagnosisin medicine, the edge of medical image contains a lot of information and it has been widelyused in image segmentation, lesions identified and virtual operation system. But in theprocess of imaging and transmission for medical image, it will often be affected by noise,and medical image relative to the ordinary natural images dose not have too muchregular edge lines, the structure is complex and the detail information is rich, sothe noise will cause great influence to the result of edge detection. In this paper, accordingto the characteristics of medical images, especially the image which is influenced by noise,we use the method of mathematical morphology to do a deep research for image edgedetection, the main contents are as follows.Firstly, the concept and classification of image edge are introduced briefly. Severalclassical edge detection algorithms are described, for example, Robert, Sobel, Canny andPrewitt. Then the performance of algorithms is analyzed through the simulation experimentof medical image and the advantages and disadvantages are pointed out.Secondly, the paper introduces the knowledge of set theory which will be used and thebasic principle of mathematical morphology. The calculation method of morphologytransformation is introduced by mathematical formula, and the visual effect of morphologytransformation is presented by graphic. Through the morphology transformation simulationexperiment of medical image, the four kinds of morphological transform denoising effectand short comings are analyzed. In view of de-noise effect of traditional open and closeoperation is not ideal, in this paper, the improved de-noise method is proposed. Instead oftraditional open and close operation using the same structural elements of the method, thispaper adopts two mutually vertical structural elements for denoising. Simulationexperiments show that the improved denoising method not only reduces two morphologicaltransformation process, and the denoising effect compared with traditional open and closeoperation more noise is reduced.Then, this paper introduces several kinds of basic gradient operator. In view of problemthe basic gradient operator is sensitive to the noise, using the improved denoising algorithmto construct an anti-noise gradient operator for image which has low noise.For extreme situation that images containing a high density of noise, the combination ofmorphological transformation is improved again and the complexity of the operator isincreased to construct a high anti-noise gradient operator. Simulation results show that theoperator can remove most noise effects and to the tiny structures in the medical image, itdoes not loss too much information so achieve the desired effect.Last, the important role of multi-form and multi-scale structural elements inmathematical morphology is introduced. For the edge detection of gradient operator which is presented in this paper, the big structure element is decomposed into linear decompositionof structural elements. For each structure element to ensure that there is a vertical structuralelement and the scale is determined according to number of decomposition. Simulationresults show that edge detection algorithm of high anti-noise based on mathematicalmorphology which is proposed in this paper can not only remove the noise but also detectcomplete and clear information of edge.