A new approach based on computer vision and its application for sperm analysis

In aquaculture, to improve the reproductive efficiency is a matter that experts seek solutions. This matter is based on the detection of factors that increase the fertilization rate in male sperm cell and in female ovary cells. One factor that increases the fertilization rate is strong and healthy sperm production. Semen analysis is performed for determination of healthy and strong sperm cells. The informations obtained by sperm analysis are; parameters such as the number of sperm, sperm density per milliliter, morphology and motility. There are three methods commonly used in sperm analysis. This methods are identifying with the conventional method, spectrophotometric systems and CASA (Computer-Assisted Semen Analysis) systems. The CASA systems, that the World Health Organization also recommends, has begun to take the place of other methods. CASA is a decision support system that aiming to reveal sperm analysis parameters with image processing techniques. CASA systems, firstly developed for human sperm cells, have been developed for other living things later. In this study, a fast and efficient CASA system, analyzing the sperm cells of Oncorhynchus Mykiss, is designed. In the study, sperm video images, taken from a video camera attached on the microscope, were analyzed in OpenCV platform using C ++ and C # .NET programming languages. Sperm count and concentration was determined by utilizing the circle Hough transform and adaptive thresholding on detection of sperm cells. To reveal the movements of sperm cell parameters, tracking multiple targets was made on the video. For Multi-target tracking, background extraction performed by VIBE algorithm, Kalman filter is used to assign labels to targets, Hungarian algorithm is used for assignment problem formed by the intersection of targets path. With the obtained informations target tracking, sperm quality parameters are calculated. For calculation of VCL and VAP, located in these quality parameters, is proposed a new approach using Lagrange interpolation. Quick and effective results in the determination of quality parameters and count of sperm cells has been shown to be obtained with developed system in this thesis. This thesis is supported by TÜBİTAK (Scientific and Technological Research Council of Turkey) in 1512-Stage Entrepreneurial Assistance Program's project as numbered of 2140052 and entitled of "Computer Aided Sperm Quality Analyzer Software Development-BASA". KEYWORDS: Sperm Analysis, CASA, Computer Vision, Circle Hough Transform, Lagrange Interpolation