一种静态背景下的运动目标检测算法研究  

A Method of Moving Object Detection Arithmetic under the Static Background

作　　者：李晋惠[1] 容慧[1] 

机构地区：[1]西安工业大学数理系,西安710032

出　　处：《西安工业大学学报》2008年第6期573-576,共4页Journal of Xi’an Technological University

基　　金：陕西省教育厅自然科学基金(03JK171)

摘　　要：运动目标检测技术是计算机视觉研究的重要课题.对常用的Kim目标检测方法,针对其不足进行了改进.将连续三帧的差分图像和背景差分图像直接相乘得到灰度图像,然后对该灰度图像进行自适应最大方差阈值分割,再基于灰度加权图像模板匹配法实现目标的检测.结果表明,该方法能够准确地检测出跟踪目标,检测信息比较完整,具有较好的实用价值.Moving object tracking is one of the important task in the filed of computer vision. In this paper, the common method-Kim object detecting is introduced. Aimed at it＇s shortage, an advanced method is put forward. The gray image is got by multiplying the adjacent three differ-fames and background differ-fame, then, the image segmented with OSTU, the gray weighted model image is utilized to match the segmented picture in order to detect the moving object. With the proof of experiment, the improved method can detect the object accurately, and the integrity of detected message is better than the Kim method. The improved method has the good value of application.

关 键 词：目标检测 Kim方法 最大类间方差阈值分割 模板 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]