Study Of The Automatic Segmentation And Counting Technique Of Wall-Pasted Cells

In vitro anti-virus experiment, precisely counting of living cells is necessary and important. The traditional methods of examining the antivirotic medicine criteria restrict the accuracy for living cells counting. And following the development of photoelectric technology and image processing technology, a new powerful technology is provided for cell counting and analysis, which is able to improve facticity of medicine screening experiment result. In this paper, we research Hela wall-pasted cells and analyze the characteristic of microscopic images of wall-pasted cells. According to the characteristic, we respectively study preprocessing, segmentation and counting . The main content is as following:(1) The research of image preprocessing. A preprocessing method for microscopic images of wall-pasted cells which can improve image quality is studied.(2) The research of image segmentation algorithms. It is an important studying content of the paper. There are three segmentations that are provided. They are the segmentation algorithm basing on edge linking, the segmentation algorithm basing on improved C-V active contour model, the segmentation algorithm basing on region growing and distance transform criterion. Analyzing advantages and disadvantages of three algorithms and estimating practical application, the algorithm basing on region growing and distance transform criterion is chosen segmentation method of counting scheme finally.(3) The research of automatic counting algorithms. Connected region labeling algorithm is mainly applied, by which small regions are eliminated.In order to test the superiority of the algorithm, microscopic images with different sizes and different qualities are used to experiment and compare. The experiment result proves the scheme is efficient and provides the high accuracy for cell counting.