基于Gabor变换的特征点跟踪算法研究  

New feature point tracking algorithm based on Gabor wavelet transformation

作　　者：耶亚林[1] 史仪凯[1] 王萑[1] 张倩[2] 

机构地区：[1]西北工业大学机电学院,西安710072 [2]西安通信学院,西安710106

出　　处：《计算机应用研究》2009年第8期3198-3200,共3页Application Research of Computers

基　　金：航空科学基金资助项目(20072E53050)

摘　　要：为了实现对一组动态特征点的跟踪,基于Gabor变换的特征点提取方法提出了一套跟踪方案。首先采用仿射变换描述点的轨迹模型,引入PDAF算法进行状态估计;然后基于Gabor变换特征向量的匹配算法进行对应点辨识;最终推导出EKF滤波器方程组,对时态信息进行滤波。仿真结果表明,该特征点跟踪算法能够在目标图像辨识率较低的情况下满足目标跟踪的精确性要求。In order to realize the tracking of a series of dynamic feature points, this paper put forward a tracking scheme based on feature point extraction. First of all, used simple affine transformation to describe the motion of feature points and employed PDAF to obtain the estimation of state vector. Then employed a matching method based on Gabor wavelet transformation to find the corresponding point. Finally cited an EKF to process the temporal information. The simulation of experiment shows that this tracking algorithm can meet the accuracy requirement of object tracking in the situation of low image resolution.

关 键 词：特征点跟踪 GABOR变换 概率数据关联滤波 卡尔曼滤波 图像匹配 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]