一种新的基于频域的相关跟踪算法  

A New Correlation Tracking Method Based on Frequency Domain

作　　者：李蓉[1] 周维柏[1] 

机构地区：[1]华南师范大学增城学院,广东广州511363

出　　处：《昆明理工大学学报（理工版）》2009年第5期36-41,共6页Journal of Kunming University of Science and Technology(Natural Science Edition)

基　　金：广东省本科高等教育教学改革项目(项目编号:BKJG200765)

摘　　要：提出了一种定位精度高、运算量小的实时相关跟踪算法.算法通过对目标进行差分尺度空间滤波,提取目标的区域特征,增强了相关匹配峰值强度和抗干扰性;其次,将大规模的相关匹配运算由时域转换到频域进行,从而保证了相关跟踪的实时性;利用加权的目标边缘质心和形心的偏差量后向抑制模板更新时的漂移量,从而减弱模板更新时的模板漂移现象,保证了复杂场景下的目标跟踪.实验表明,算法能够适应在低对比度、复杂场景下的目标跟踪,同时对跟踪过程中摄像头抖动也不敏感.仿真结果表明,该算法运算耗时低,配准峰值强.A new correlating tracking algorithm of low computation and high precision is presented in this paper. The algorithm firstly extracts the region features from the scale filters, which intensifies the corresponding peak values and the anti - interference. It then does high computation in frequency, which ensures the real - time tracking. Finally,it updates the template using the difference between the centroid and center,which can well overcome the drifting problem. It is shown through the experiments that this method ensures good tracking performance in low contrast and complex background and that it can adapt to the camera shaking. The simulation resuits show that the algorithm is low time -consuming and strong quasi -peak.

关 键 词：相关跟踪 尺度空间 模板漂移 频域 质心 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]