基于HSV颜色特征的多目标视频检测与跟踪算法

摘要

Aiming at the problem that the target is easily lost when traversing multiple moving targets, an algorithm based on HSV color space to identify and track different color targets is proposed, which is robust to illumination.After the median and Gaussian filtering, the filtered binary image is subjected to morphological dilation by brightness detection, then the edge image of the target binary image is detected to obtain the minimum value of the original binary image.Enclosing the rectangle frame, drawing the motion trajectory corresponding to each color target after detecting frame by frame, and completing the multi-target video tracking.Compared with the classical Cam-Shift algorithm, the experiments show that when multiple targets have different color features, the problem of multi-target interlaced motion lost targets and sensitive to illumination in video tracking is solved effectively, and the tracking precision is high.%针对跟踪多个运动目标交错运动时容易丢失目标的问题,提出了一种基于HSV颜色空间去识别与跟踪不同颜色目标的算法,对光照具有一定的鲁棒性.通过颜色空间转换提取目标原始二值图像,经中值和高斯滤波后,通过亮度检测判断是否对滤波后二值图像进行形态学膨胀;之后对较为精确的目标二值图像进行边缘检测,得到其最小包围矩形框,逐帧检测后绘制各颜色目标对应的运动轨迹,完成多目标的视频跟踪.对比经典Cam-Shift算法,实验表明,在多个目标具有不同颜色特征时,有效解决了视频跟踪中多目标交错运动丢失目标,且对光照敏感的问题,跟踪精度较高.