J4 ›› 2009, Vol. 27 ›› Issue (06): 652-.

• 论文 • 上一篇    

几种人体运动检测算法的比较研究

周 游1|刘艳滢1|王春民2|陈 静1|陈秋萍1,魏雅娟1   

  1. 1.中国科学院  |长春光学精密机械与物理研究所|长春 130033;2.吉林大学 |仪器科学与电气工程学院|长春 130061
  • 出版日期:2009-11-20 发布日期:2009-12-18
  • 通讯作者: 周游(1986— ),男,江西萍乡人,长春光学精密机械与物理研究所硕士研究生,主要从事图像压缩、图像检 测研究, E-mail:aniuxhs@163.com
  • 作者简介:周游(1986— )|男|江西萍乡人|长春光学精密机械与物理研究所硕士研究生|主要从事图像压缩、图像检 测研究,(Tel)86-13944828200(E-mail) aniuxhs@163.com;刘艳滢(1965— )|女|辽宁营口人|长春光学精密机械与物理研究所研究员,硕士生导师|主要从事数字图像处理研究| (Tel)86-13654319757(E-mail)liuyy@ciomp.ac.cn

Contrast Research of Several Human Motion Detection Algorithm

ZHOU You1|LIU Yan-ying1|WANG Chun-min2|CHEN Jing1|CHEN Qiu-ping1,WEI Ya-juan1
  

  1. 1.Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Scienes,Changchun 100033,China;2.College of Instrumentation and Electrical Engineering, Jilin University, Changchun 130061,China
  • Online:2009-11-20 Published:2009-12-18

摘要:

为了提高计算机视觉运动人体检测的效果,采用了OGHMs(Orthogonal Gaussian-Hermite Moments)法应用于人体运动检测中,在实现过程中对OGHMs加以简化,对运动图像进行Gaussian平滑处理,并在模块运算后期引入腐蚀运算从而得到更精确的检测效果。分析了时间差分法、背景减除法、光流法等常用的几种运动人体检测算法。通过对现有的实验结果分析比较表明,增加腐蚀作用后的OGHMs法不仅具备常用算法各自具有的一些优点,而且抗干扰能力强,同时还较好解决了计算量大的问题,不需要进行背景重建和更新等步骤,并且检测出的运动信息较多,能较完整、较精确地检测出运动人体。

关键词: 人体运动检测, 正交高斯Hemite矩, 时间差分法, 背景减除法, 光流法

Abstract:

In order to improve the effect of the Moving human detection in computer vision, this paper used the OGHMs method. In the implementation process, we have simplified OGHMs(Orthogonal Gaussian-Hermite Moments),used the Gaussian smoothing to the movement image, and added an erode algorithm to get more accurate results. Simultaneously several methods used in moving human detection are analyzed, such as temporal difference, background subtraction and optical flow. The experimental result indicated that the OGHMs which is added erode process has some advantage of common algorithms, and has strong anti-jamming capability. This algorithm has better solved the problem of large quantity of calculaion, it is not necessary to carry on the process of background rebuilding and updating. The detection result of this algorithm contains more movement information, it can finish the moving human detection more integrallty and accurately.

Key words: moving human detection, Orthogonal Gaussian-Hermite Moments(OGHMs), temporal difference, background subtraction, optical flow  

中图分类号: 

  • TP751