结合小波变换与相位相关的图像匹配方法  

Image Matching Method Using Wavelet Transform and Phase Correlation Technique

作　　者：邱庆军[1] 徐可欣[1] 蒋景英[2] 虞启琏[2] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072 [2]天津大学精密仪器与光电子工程学院,天津300072

出　　处：《信号处理》2003年第5期469-472,共4页Journal of Signal Processing

基　　金：感谢国家自然科学基金委对精密测试技术及仪器国家重点实验室的资助;项目编号30170261

摘　　要：图像匹配技术是图像测量系统中的关键技术,是目标定位中最基本的方法。相位相关作为一种图像匹配技术对噪声有较高的容忍程度,检测结果与照度无关,受几何失真影响小。但该方法对旋转变化非常敏感。为解决此问题,本文提出结合小波变换和相位相关的匹配算法,该算法首先对图像进行小波分解,并对结果中的近似信号进行空域旋转校正;然后利用相位相关方法进行图像匹配,从而达到快速精确的定位。该方法在掌纹图像匹配定位中的应用证明了它的有效性。Image matching, the key technique of image measurement system, is a fundamental method of target location. Phase correlation algorithm as a kind of image matching method has many advantages such as insensitive to noise, unrelated to illumination of target, immune to geometric distortion. However, the algorithm is very sensitive to the rotation of target image. To eliminate the influence, this paper provides an efficient matching algorithm, which combines wavelet transform (WT) with phase correlation. The algorithm decomposes the target image with wavelet transform, and then the rotational compensation is performed with the approximate signal, finally the fast and accurate matching is accomplished by using the phase correlation algorithm. The present algorithm has been applied to the location system based on palm-print image matching, and the result shows its validity and accuracy.

关 键 词：图像匹配 小波变换 相位 图像模式 图像测量系统 图像处理 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]