Research On The Improvement Of Multi-Targets Tracking Of Sperm Motility Video Based On Multi-Threshold Segmentation

With the development of the computer technology, the computer-aided sperm analysissystem based on the image processing technology is used successfully in clinical trials, andgradually replace the traditional methods of observing the sperm counts by the naked eyes,judging the sperm motility by the rule of thumb and analyzing and diagnosing invalidly, thusprovides a kind of reliable and advanced technology for the modern reproductive Hospital’sdiagnosis of male infertility and artificial insemination sample selection.Nowadays, Domestic computer-assisted semen analysis (CASA) system is all based onthe microscope image video, the image segmentation of the sperm heads’ video frame, andthe multi-objective tracking of the identifier of the sperm head. Then we can automaticallycalculate the trajectory of the sperm to get the parameters of the sperm motility. Although theexhibition effect is good, but there are still some shortcomings, such as the target tracking hascertain probability to lose, the accuracy of Analysis results is susceptible tothe sample solubility, the sample processing robustness is low and etc.This article is based on the comprehensive analysis of the present computer-aided spermanalysis technology, we expand the single-threshold segmentation, and use themulti-threshold segmentation algorithm to extract our two goals: the sperm head and thesperm tail, also we optimize the time-consuming multi-threshold segmentation algorithm andmake use of the information extracted by using the multi-threshold segmentation to the spermimage to start multi-target tracking. Combining the information of the sperm tail and thesperm procession dynamics theory, we judge the possible direction of the movement of sperm,then by using the information of the direction of the movement of sperm to model the spermmotion, we can reduce the search scope of the sperm target tracking, and therefore improvethe accuracy and efficiency of multi-target tracking. Finally, in the matching process of thesperm motion tracking, we add more characteristics of the sperm, such as the area of thesperm tail, and its length and direction in order to get more methods of the match of sperm,and also make the match effect of the sperm better. Through the multi-threshold segmentationand the improved multi-target match tracking, we achieved the improvement of thealgorithm’s performance and the improvement of algorithm’s performance is proved byexperiments.