摘要
提出了一套检测和识别视频序列图像中运动物体的算法.该方法在连续3帧视频图像中,分别在第1和第2帧之间、第2和第3帧之间作差分运算,根据未变化区域与运动变化区域服从不同的统计规律设计检测门限,对差分图像作运动变化检测和连通域的识别.将检测到的运动变化图像作去噪和空域分割后,再对得到的两幅差分图像作相与运算以确定运动图像的位置,最后再基于运动物体自身的灰度信息,恢复出完整的运动图像.该算法适于应用在实时视频监控场合,实验结果令人满意.
An approach for detection and recognition of moving objects in sequential images is presented. In this method, the difference operations among three sequential frames are conducted firstly between the former frames and the latter ones in turn. Then the difference images are used for determining the threshold according to the different distribution between the moving image and the unmoving part, and for moving detection and spatial segmentation. After that, the location of the moving object in the images is determined by a logic and operation. At last an image retrieval algorithm is applied to obtaining the complete moving image by using the gray information of the moving image itself. The validity of the method is confirmed by experimental results given in this paper.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2004年第1期122-126,共5页
Journal of Dalian University of Technology
关键词
运动图像
图象检测
图象识别
运动目标提取
视频监视
video monitoring
moving image detection
detection and retrieval of moving objects
