精子多目标检测和跟踪算法的设计与实现  

Design and implementation of algorithm of multiple detecting and tracking for sperm analysis

作　　者：于江有[1] 王知衍[1] 张艳青[1] 

机构地区：[1]华南理工大学计算机科学与工程学院,广东广州510006

出　　处：《计算机工程与设计》2010年第9期2076-2079,共4页Computer Engineering and Design

摘　　要：针对精子检测和计算机辅助的精子运动特征分析所存在的问题,分析了精子密度大、运动速度快等特点和实际应用中遇到的各种困难,比较了多种阈值分割和多目标跟踪的方法,设计了基于矩量保持和卡尔曼滤波的精子检测与跟踪算法。利用矩量保持法对精子图像进行阈值分割和卡尔曼滤波对精子运动目标进行预测。实验结果表明,该算法在检测时间和精确度上都有很大提高。Aimed at the problems of the computer-aided sperm analysis （CASA）, sperm density, movement speed, etc are analyzed, a wide range of threshold segmentation and multi-target tracking methods are compared, the moment-preserving and Kalman Filter key technologies are analyzed, and an efficient and low-cost method for automatically detecting and tracking the moving object from sperm image sequences is presented. The algorithm improves the effectiveness and accuracy of sperm detecting.

关 键 词：目标检测 多目标跟踪 矩量保持 卡尔曼滤波 计算机辅助精子分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]