摘要
首先采用平均背景法对混合高斯建模获得的背景进行处理得到纯净的背景,然后在当前图像与背景图像之间进行差分运算,最后通过轮廓筛选的方法去除目标图像中的干扰点,并通过形态学处理对目标中的间隙进行填充,获得完整目标。实验结果表明,对于运动目标及停留的目标,该算法的识别率和误检率分别为86.7%和8.3%。
The average background method is used to deal the background obtained from mixed Gaussian to purify the image background.Then the difference between current and background image is calculated,and the interference points in the image are removed with contour screening.The gap in the image is filled with morphological processing to obtain the final image.Experimental results indicate that the recognition rate and false rate are 86.7%and 8.3%respectively,for both the moving and staying target.
出处
《长春工业大学学报》
CAS
2017年第5期421-425,共5页
Journal of Changchun University of Technology
基金
吉林省教育厅"十三五"科学技术研究基金资助项目(2016-35)
关键词
目标检测
混合高斯模型
平均背景
轮廓筛选
target detection
mixed Gaussian model
average background
contour screening
作者简介
刘丽伟（1974-）,女,汉族,吉林通化人,长春工业大学副教授,博士,主要从事模式识别和智能系统方向研究,E-mail：liuliwei@ccut.edu.cn.
