基于机器视觉的运动目标检测方法研究  

Research on Method of Detecting Moving Object with Machine Vision

作　　者：张燕超[1] 徐桂云[1] 崔吉[1] 段松杰[1] 

机构地区：[1]中国矿业大学机电工程学院,江苏徐州221008

出　　处：《煤矿机械》2008年第3期65-66,共2页Coal Mine Machinery

基　　金：江苏省博士后科研基金资助项目(0502010B);中国矿业大学科技基金资助项目(2005B005);江苏省"六大人才高峰"项目(06-E-052)

摘　　要：针对复杂背景下机器视觉系统中运动目标的检测,提出一种基于差分图像的列灰度均值分布的运动目标检测方法,该方法以自适应更新法建立背景模型,根据差分图像的列灰度均值分布图来实现差分图像的二值化及二值差分图像中阴影和噪声处理。实验表明,该方法有效去除了运动目标阴影及背景噪声,可准确地检测出运动目标。Aiming at the problem of detecting moving object from complex background with machine vision system, a method of detecting moving object using the distribution of column gray level mean value of difference image is put forward. With this method, the background model was created using adaptive update, the binarization of difference image was implemented and the noises and shadow were removed according to the distribution map of column gray level mean value of difference image. The experiment shows that, with the method the shadow of moving object and noises can be removed from background effectually and moving object can be detected exactly.

关 键 词：机器视觉 运动目标检测 列灰度均值分布 自适应更新 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]