摘要
针对视频序列中运动目标的检测与跟踪进行了研究,提出了一种基于视频序列的YCbCr模式归一化定义,利用三帧图像差分,区域生长生成运动矩形框的方法对目标进行运动分割;并利用基于速度差的粒子滤波器对运动框的移动进行预测,从而达到快速跟踪运动目标的目的。实验结果表明通过调节内部参数,在固定摄像机不复杂背景下对人、车运动目标检测与跟踪是快速且有效的。
Aimed at researching on detecting and tracking of moving objects in video- sequences, an algorithm is proposed in this paper to carry one＇s point to track moving objects expeditiously. Firstly, The video -sequences have been defined based on YCbCr color space. Secondly, use of three frames difference and region growing to moving segmentation by moving presenting rectangle frames. At last, a particle filter based on rate difference is used to forecast the moving of the moving rectangle frames. Experimental results prove that the algorithm is simple and effective. By adjusting several interior parameters, it＇ s rapid and effective to detect and track human and vehicle fulfilling the qualifications for fixed video and common surroundings.
出处
《仪器仪表用户》
2008年第2期8-10,共3页
Instrumentation
关键词
三帧差分
块匹配
粒子滤波器
three frames difference
block-matching
particle filter
