摘要
针对目前运动目标的跟踪与记录方法占用存储空间较大的缺点,提出了一种减少存储空间的记录算法,即先用三帧差分算法和Snake算法相结合检出运动物体的轮廓,再利用Hausdorff算法对提出的轮廓进行匹配,并将匹配后的轮廓和运动轨迹以文本文件存储,大大降低了运动目标轨迹记录存储容量.实际运用表明,改进后的记录存储空间相当于通常视频文件的万分之一.该算法适于长时间记录运动目标轨迹.
At present, data from tracking moving targets consumes too much memory. An improved recording algorithm is proposed that reduces memory requirements. First, initial points are obtained by using a three-frame difference algorithm. Then the Snake algorithm is used to extend the initial points to determine the object＇s contour. Finally, the Hausdorff distance measurement matches contours to standard figures and the results are saved in a text file. The improved algorithm uses approximately 0. 01% of the usu al memory required, so it is suitable for storing data from long-term tracking.
出处
《智能系统学报》
2008年第2期145-149,共5页
CAAI Transactions on Intelligent Systems
基金
吉林省科技发展计划资助项目(20060531)
