基于局部背景特征点的目标定位和跟踪-中南大学学报.PDFVIP

第 47 卷第 9 期 中南大学学报( 自然科学版) Vol.47 No.9 2016 年 9 月 Journal of Central South University (Science and Technology) Sep. 2016 DOI: 10.11817/j.issn.1672-7207.2016.09.018 基于局部背景特征点的目标定位和跟踪 张铁，马琼雄 (华南理工大学 机械与汽车工程学院，广东 广州，510641) 摘要：针对存在多个与目标相似的区域以及摄像头经常运动的跟踪问题，提出一种基于局部背景特征点的目标定 位和跟踪方法。首先，根据相邻匹配的局部背景特征点与目标的位置关系对目标进行定位，然后，以定位得到的 预测位置作为搜索起点，结合粒子滤波和均值偏移方法获取目标的候选区域，最后，根据候选区域和预测位置的 距离对候选区域的相似度加权，将加权后相似度最高的候选区域作为跟踪结果。研究结果表明：该算法能够避免 周围相似物体的干扰并准确跟踪目标，具有较好的鲁棒性和实时性。 关键词：粒子滤波；均值偏移；局部背景特征点；目标定位；目标跟踪 中图分类号：TP391 文献标志码：A 文章编号：1672−7207(2016)09−3040−10 Target location and tracking based on local background feature points ZHANG Tie, MA Qiongxiong (School of Mechanical Automotive Engineering, South China University of Technology, Guangzhou 510641, China) Abstract: To solve the problems of multiple similar objects and camera motion in target tracking, a method of target location and tracking was proposed based on local background feature points. Firstly, a target position was predicted by considering the position relation between the target and the matched feature points surround target of previous image. Then candidate objects were searched starting from the predicted position by combining particle filter and mean shift. Finally, the similarity of each candidate object was weighed by the distance between candidate object position and predicted position. The final target was the candidate object with the highest similarity. The results show that the proposed algorithm can track targets acc