The Research Of Brine Leucorrhea Cells Microscopic Image Recognition’s Critical Technology

With the rapid development of science and technology computer technology has been used medical diagnosis more and more widely, and one of the most important application is to recognize and classify cell micro-images by identification technique in medicine field, the key technologies of brine leucorrhea cell images will be studied in this paper.Leucorrhea routine examination is one of the most widely examinations in gynecological diseases, by observing the number of bacillus, micrococcus and white cells in micro-images to judge the cleaning in order to determine whether there is inflammation or not. A variety of brine leucorrhea cells and its ingredient already make the image segmentation difficult. Weak contour, overlapping, tininess, impurity compounded the difficulty. In this paper, we will combine computer technology and biological medicine to recognize brine leucorrhea cell automatically in order to improve clinical diagnostic efficiency.The paper use digital image processing and pattern recognition technology to achieve the segmentation and recognition of brine leucorrhea cell automatically, and research a series of algorithm including image segmentation, feature extraction and recognition. The point is image segmentation and feature extraction, the feature parameters will be used to cell recognition, calculation number and the judgment of cleaning.In order to segment image accurately, we research several segmentation methods such as threshold segmentation, segmentation based on edge extraction and region firstly. We will introduce the level-set method which will be more suitable for people vision, and then improve the level-set method to solve the problem that the traditional level set method fails to work. The tests show that the segmentation effect is very good.After the image segmentation we will calculate the geometric characteristics and shape characteristics such as perimeter, area, circular degree and rectangular degree, etc. In order to recognize clue cells and epithelial cells we need to extract their texture feature such as mean, smoothness, entropy, etc. And these features will be used as the basic of cell classification.The cleaning has four types, Ⅰ degree, Ⅱ degree, Ⅲ degree and Ⅳ degree. we will use the features of cells to classify and recognize these cells. The most important is that we need to recognize the bacillus, micrococcus, epithelial cells and white cells, then we will judge the cleaning according to the four criterion by the number of these cells.