摘要
针对全局运动视频序列中的目标检测,提出了改进的灰度投影算法。通过对灰度投影相关曲线的分析,根据其单峰性特征,提出了三点局域自适应搜索算法,该方法能够快速的估计出前后两帧之间的运动矢量。然后以参考帧图像背景为参照,映射当前帧的背景信息,以此补偿全局运动矢量,将连续几帧图像的相同背景稳定在同一幅图像的相同位置上,从而能够利用改进的连续三帧差分法准确地检测出运动目标,并根据目标的特征进行分类识别。实验结果表明,该方法能够有效地从视频序列中提取和识别出运动目标。
The improved gray scale projection algorithm is proposed based on the moving objects detection of video sequences with global motion. Through analyzing the related curves, three-point locally adaptive searching algorithm is proposed on the basis of single peak characteristics, The motion vectors between the successive image frames can be estimated by using this algorithm. Through maping the background of the current image frame based on the background of the reference image frame, the global motion vectors can be compensated. The background of successive image frames are fixed to the same position of the same image frame. Then the moving objects can be detected exactly by using a method of improved three successive image frames subtraction and classed with their characteristics. Experimental results demonstrate that this method can effectively extract and distinguish moving objects with changing background.
出处
《计算机与数字工程》
2008年第10期165-169,共5页
Computer & Digital Engineering
关键词
目标检测
灰度投影
运动补偿
三帧差分
object detection, gray scale projection, move compensation, three image frames subtraction
