Research On The Estimation Of The Number Of Red Blood Cells Based On Image Processing

There are about five million red blood cells in adults,and the number of red blood cells can be used for disease detection and judgment.The method for detecting the number of red blood cells is instrumental and manual detection.However,the instrument method is expensive,and the manual detection method is time consuming and laborious.In response to these problems,digital image processing has become the focus of the research due to its key role and outstanding performance in cell analysis.Digital image processing is not only low cost,but also reliable.Therefore,this thesis proposes an automatic treatment method for human red blood cell counting using the image processing method.In order to realize the automatic counting of red blood cells,the automated image processing counting system proposed by the thesis consists of three parts: cell image preprocessing,cell segmentation and counting.The work and contributions of the thesis are:(1)A quick and convenient treatment method for red blood cell counting is proposed,and specific implementation steps of image processing are mainly performed,pretreatment,red blood cell contour extraction and segmentation,overlapping red blood cell segmentation and counting results.The preprocessing is mainly to remove the image noise and the background illumination correction.The red blood cell contour extraction and segmentation is to separate the background from the background,and then the re-segmentation of the overlapping red blood cells is performed for the accuracy of counting.(2)In the aspect of red blood cell contour extraction and segmentation,the various contour extraction and segmentation algorithms used are mainly studied,including edge detection,threshold segmentation,boundary tracking and other algorithms.These algorithms are also used for red blood cell image contours.The effects of extraction and segmentation are compared.(3)In the aspect of segmentation of overlapping red blood cells,the morphological method,the watershed segmentation method and the chain code method are mainly studied,and their effects on overlapping red blood cell segmentation are compared,and the advantages and disadvantages are summarized.(4)On the basis of studying the effect of existing algorithm on contour extraction and overlapping segmentation of red blood cell images,the thesis proposes to use the combination of adaptive segmentation algorithm and pit segmentation method to segment red blood cell images and accurately segment them.The rate and red blood cell count results were analyzed and compared.