基于阈值判断的CamShift目标跟踪算法

针对CamShift算法只利用目标的颜色信息,在跟踪过程中,易受目标相似物、遮挡以及光照等复杂背景影响导致目标搜索窗口发散,跟踪稳定性能降低,提出了一种基于阈值判断的目标跟踪方法。该方法将OTSU法和Snake模型结合,利用OTSU法以最佳阈值对图像进行分割,分离前景区域和背景区域,初步提取目标轮廓作为Snake模型的初始轮廓,经收敛得到目标的精准轮廓,利用轮廓外接最小矩形框内的像素计算目标质心,判断与CamShift算法中目标搜索窗口质心之间的欧式距离,如果未超出阈值,则直接使用CamShift算法跟踪目标,反之,则将计算出的目标质心作为CamShift算法中当前帧目标搜索窗口的质心跟踪目标。实验结果表明,该算法跟踪目标具有较好的实时性,跟踪性能稳定、可靠。

Moving target detection and tracking algorithm based on contour and ASIFT feature matching

In the process of target tracking, the CamShift algorithm only uses the color information of the target to achive tracking, and the complex background such as similar objects, occlusion, the interference of light and so on, which would lead to the divergence of the target search window and reduce tracking performance. In order to solve the defect of the CamShift algorithm, a method of target tracking based on threshold value judgement was proposed. The method combined the OTSU algorithm with the Snake model which used OTSU to segment the image with the best threshold to separate the foreground and background, then the initial contour of the target was extracted that was taken as the input contour of the Snake model, and the precise contour of the target would generate. And the target centroid could be calculated by using the pixels in the minimum rectangle region that outsided the precise contour. The Euclidean distance between the centroid of the target search window in the CamShift algorithm and the new centroid could be as the basis for tracking. If the Euclidean distance within the set threshold, the CamShift algorithm would be used to track the target directly; On the contrary, the new target centroid was regarded as the centroid of the target search window in the current frame to achieve target tracking through the CamShift algorithm. The experimental results showed that the new algorithm had good real-time performance, and the tracking performance was stable and reliable.