小波域基于聚类分割的运动目标检测算法  

Moving objects detection algorithm based on clustering segmentation in wavelet domain

作　　者：刘英霞[1] 常发亮[2] 

机构地区：[1]山东传媒职业学院,济南250200 [2]山东大学控制科学与工程学院,济南250061

出　　处：《计算机工程与应用》2013年第18期15-19,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.60975025;No.61273277);山东省高等学校科技计划项目(No.J11LG85;No.J12LN89);教育部留学回国人员科研启动基金资助项目(No.20101174)

摘　　要：针对监控系统中目标检测精准性和实时性不能很好兼顾的问题,研究了小波域基于聚类分割算法的目标检测系统,在小波域分别给出了高频成分和低频成分自适应阈值的确定方法,推导出了判决准则和最佳阈值。利用该算法进行目标识别,可以去除系统噪声,快速检测出目标,并能保证系统的精准性。通过仿真实验对该算法进行验证,结果表明,该算法能够准确检测出目标,算法速度快,能保证系统的实时性。Because of the contradiction between accuracy and real-time in monitoring system,the object detecting system in wavelet domain based on clustering segmentation algorithm is researched.A method to determine the adaptive threshold for high and low frequency components is presented.The judging criterion and the optimum threshold are deduced.The algorithm can remove the system noise,detect the object fast,and ensure the precision of system.It is proved by simulation and experiment results that the algorithm can detect the object accurately and realize a real-time system.

关 键 词：目标检测 小波域 聚类分割 最佳阈值 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]