结合SURF与Kalman滤波的CAMShift跟踪算法

发布时间：2018-03-14 17:32

本文选题：目标跟踪　切入点：CAMShift算法　出处：《电子测量与仪器学报》2017年03期 　论文类型：期刊论文

【摘要】：针对传统的CAMShift目标跟踪算法,在出现颜色干扰,遮挡等复杂背景中容易跟丢的问题,提出了一种结合SURF特征匹配与Kalman滤波的CAMShift跟踪算法。该算法利用CAMShift算法跟踪得到的候选目标与模板目标的色度和梯度方向的综合直方图比较计算得到的Bhattacharyya系数作为判定依据,当系数大于给定阈值时,采用SURF算法对搜索窗口和上一帧跟踪结果进行特征匹配,重新计算目标的大小和位置。同时为了避免目标快速运动时跟踪失败和减少SURF匹配的计算量,利用Kalman滤波对运动目标窗口进行预测更新以确定下一帧搜索窗口的中心位置。实验表明,该算法在图像背景复杂,出现颜色干扰以及部分遮挡时能够稳定跟踪,其跟踪速度与结合SURF的CAMShift算法相比有显著提高。
[Abstract]:For the traditional CAMShift target tracking algorithm, it is easy to be lost in the complex background, such as color interference, occlusion and so on. A CAMShift tracking algorithm based on SURF feature matching and Kalman filtering is proposed. The algorithm uses the Bhattacharyya coefficients calculated by comparing the chroma of candidate target and template target and gradient direction by CAMShift algorithm. When the coefficient is larger than a given threshold value, the SURF algorithm is used to match the search window and the previous frame tracking results to recalculate the size and position of the target. The Kalman filter is used to predict and update the moving target window to determine the center position of the next frame search window. Experiments show that the algorithm can track the image stably when the background is complex, the color interference and partial occlusion appear. Compared with the CAMShift algorithm combined with SURF, the tracking speed is significantly improved.
【作者单位】：江南大学物联网学院;
【基金】：国家自然科学基金(61374047) 江苏省产学研创新项目基金(BY2014023-36,BY2014023-25)资助项目
【分类号】：TP391.41

【相似文献】