Color Image Segmentation Algorithm Research Based On Tissue Cells

Research on the algorithm of image segmentation is the first step for identification andanalysis of image segmentation algorithm selection, it will directly affect the processingresults of the image processing. So in the image processing and analysis in the course of thestudy, the optimal image segmentation algorithm improvement is becoming the main objectof study by researchers. The color tissue cells are converted into gray image containsinformation relatively, but can not get the complete segmentation results of color image. Inorder to find the optimal segmentation algorithm, ultimately color image of tissue cells areselected as the research background. Then the color characteristic properties of tissue cellimage, i.e. the contour features, image organization cell color distribution to selectsegmentation algorithm, are used to achieve the disease diagnosis, prevention and researchpurposes.According to the collected characteristics of color tissue cell image, HSV color space isfinally chosen through the algorithm comparison. Tissue cell image background containinglarge noise needs to preprocess, the first step is to process image with equalization, smoothing,sharpening and enhancing, and then selects the image filtering method, the choice offrequency filtering algorithm combing bilateral filtering and wavelet filtering is adopted in thefiltering stage. A kind of watershed algorithm combined wavelet transform and mathematicalmorphology reconstruction is applied in color tissue cells by discussing a variety of colorimage segmentation methods and comparing the advantages and disadvantages of eachalgorithm. The improved wavelet algorithm, which has the characteristic of the windowchanges with the frequency, is used to filter the image. The opening and closingreconstruction algorithm in the morphology is adopted to remove the image of small spots,modify the image of the local maxima, and realize the denoising and smoothing function;finally, the improved mark two watershed segmentation algorithm is used to label theforeground and background and segment the image.The algorithm used in this thesis has an obvious advantage in the application of tissuesections of cell segmentation through the results of Matlab simulation analysis. The boundaryof color tissue cells segmented is particularly evident, and the noise of the image is removedwell. This algorithm has the advantages of edge higher positioning accuracy and edge goodcontinuity, avoids over-segmentation, can inhibit the edge of dead cells and gets the real edgemostly at the same time.