基于背景重建的多运动目标检测与跟踪

Multi-Object Detection and Tracking Based on Background Reconstruction

为了满足多目标跟踪的快速性和准确性,以平均亮度和图像能量为约束,实现了基于块的快速背景重建。通过背景消减法、形态学滤波、区域生长,实现了多目标的检测;通过对不同的目标贴上标签,实现了目标快速匹配;然后根据相邻帧间同一目标形心位置以及外接矩形变化不大,建立目标链,实现多目标跟踪。实验结果表明,该方法具有良好的快速性和准确性。

To satisfy the goal of fast speed and high precision, and with the average intensity and image energy as the constraints, fast background reconstruction has been achieved, and the achievement of multi-object detection is done by background subtraction, morphologic filtering and region growing; the object fast matching is achieved via the labelled different objectives; and then,since there is no significant change in the centroid coordinate and outer rectangular of the same objects,it is necessary to set up objective chain so as to realize the tracking. Some experimental results show that the method has the merits of fast speed and high precision.