Research Of Segmentation Algorithms In Color Cell Images

The technique of cell detection combined computer science is moreimportant along with the development of the graphic management in medicalscience. As the significant step of cell detection, in area of cell segmentationtechnology, we need to consider the matter how to segment the cell and internalcharacters of cells by reasonable and efficient measure which exact classifyingand count for after step.This issue proposes the blue print contained blood cell pretreatment, the firstsegmentation, the second segmentation, and proposed the adaptive improvementof the algorithm in each phase of the process.First of all, we present the self-adaptive double threshold enhancedalgorithm. This algorithm has excellently enhanced effects of images byaccumulating the frequency histogram of three color weights R、G、B respectively,then getting the maximum and the minimum by the comparing with the initialthreshold. Finally, stretch the image in the range.Secondly, present the advanced FCM algorithm in hue histogram addedwindows. The new advanced FCM algorithm advances the speed of operation andthe effect of segmentation by getting the main hue area via hue histogram addedwindow, to reduce the participated pixels of FCM algorithm.Finally, propose the advanced FCM algorithm combined the globalinformation and advanced area growing algorithm with hue statistical characters.The advanced FCM algorithm improve the effect by getting the hue characters,class area and initialization via area growing of whole image, and changing thesubordinate function. The advanced algorithm enhances both the timeliness andaccuracy by analyzing the hue histogram to get the seed and growing criterion.Above all, this article presents the system of cell segmentation and advancedalgorithms of each phases for better adaptively segmentation of blood cellimages.