一种实时的鱼类轨迹跟踪算法  

A Real-time Trajectory Tracking Algorithm of Fish

作　　者：田超[1] 黄志勇[1] 

机构地区：[1]三峡大学计算机与信息学院,湖北宜昌443002

出　　处：《信息通信》2017年第1期47-50,共4页Information & Communications

摘　　要：鱼类行为学数据是鱼类研究的基础信息。计算机视觉技术是提取鱼类行为学数据的重要途径之一。文中提出了一种基于前景检测跟踪鱼类游动轨迹的方法,使用VIBE前景检测的方法将鱼类与背景区分开,再用区域搜索的方法得到鱼类游动轨迹。实验得到了鱼的游动轨迹。通过测试15000帧的视频,204帧丢失,丢失率1.36%。目标丢失后该方法能够快速找回。该方法准确稳定,可作为鱼类研究的基础数据提取手段。Fish behavioral data is the basic information for fish research.Computer vision technology is one of the mostimportantways to extract data of fish behavior. A method of Tracking Fish＇s Trajectory based on Foreground Detection is proposed, whichuse the VIBE foreground detection algorthm to separate the fish from the background. Region search method is proposed to trackthe fish trajectory from the binary image. In the experiment, The swimming trajectories of the fish were obtained.After testing15,000 frames of video, 204 frames were lost, with a loss rate of 1.36%.The method can be quickly retrieved after the target islost.The tracking method is accurate and efficient, and can be used as a basic data extraction method for fish research.

关 键 词：鱼类 视频跟踪 前景检测算法 区域搜索 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]