一种抗遮挡的多运动目标跟踪改进算法

Improved Anti-occlusion Tracking Algorithm of Multiple Moving Object

由于传统空间欧氏距离最短法难以解决遮挡问题,提出一种固定场景下抗遮挡的对多个运动目标进行实时检测和跟踪的算法.在分析传统帧差算法的优缺点的基础上对其进行了改进,引入空间滤波和区域填充.介绍了传统空间欧氏距离最短法,分析了它的缺点.用带状态参量的空间欧氏距离最短法对每个视频运动目标质心进行关联,监测每个视频运动对象的运动状态、运动轨迹.通过实验证明,该方法在改进传统欧式距离最短法的基础上,能实时有效得跟踪运动目标.

As traditional minimal Euclidean distance can not resolve problem caused by occlusion, this article introduces an anti-occlu- sion method that can detect anti track moving objects in one fixed scenario at real-time. Improving traditional coterminous frames differ- encing by add in region filling and coefficient spatial filter based on the analysis of it. By using the method which based on the minimal dimensional Euclidean distance with the status parameters, it conjuncts every centmid coordinates of the moving object and monitors the movement condition and the trajectory of each video moving object. Through the experiments and the results, it proves that the new method is more efficient and has more advantages than the traditional Euclidean dimensional minimal distance method before.