The Study Of Image Segmentation Technology Based On Mathematical Morphology Of Sections Of Tissue Cells

With the rapid development of science technology, technology of medical cell imageprocessing has been greatly improved. In the study of biological cell images, mainly on themorphology of the cells are identified and extracted. Technology of cell image segmentationimproves has great significance to realize the3D reconstruction of cell microscopic andultramicroscopic structure, to provide more favorable information for the correct diagnosis ofdiseases, to promote the cell information from quantitative to qualitative analysis andprocessing, and the study of cell mutation.Tissue cell image segmentation is to separate the cells from background of the image, forprocessing a single cell. This paper first collected cell image of color tissue sections grayprocessing, separating cells base on the gray image segmentation. The gray image ispreprocessed, correcting uneven illumination by morphological of bot-hat transform;enhancing cells edge details by top-hat transform; and adjust the image contrast, thedifference is of gray scale pulled big between cells and background. Then the image of thepreprocessed is segmented using OTSU method, the cells are all separation from background;By setting the area threshold removing impurities; filling image holes, smoothing theboundary of target region by using morphological open operation, removing burrs on theboundary. For image segmentation of adhesion cell, first the binary image is distancetransformed, which is processing by morphological open operation to obtain the distance map,eliminating the spurious local maxima, finally watershed transform based on the new distancemap and get the boundary line of single cells, and cells are all divided apart from images.Through the segmented image of tissue cell, cell counting statistics can be carried out, and thefeature of area, circularity, the external can be extracted and analyzed.The segmentation algorithm directly affects the subsequent analysis of morphology ofthe tissue cell. In this experiment, the algorithm realizes the separation of cells frombackground in the image segmentation, and the adhesion of cells were correctly separated, theobtained edge of cells can be well compared with gray images, better than other segmentationalgorithm on high accuracy and adaptability, reached the expected goal.