基于相位相关法的实时模板定位方法  

Real-time and robust pattern localization for phase correlation

作　　者：刘晨[1,2] 刘后标 贾可[1] 孙晓刚[1] 

机构地区：[1]中国科学院成都计算机应用研究所,成都610041 [2]中国科学院大学100049

出　　处：《计算机应用》2015年第A01期252-254,266,共4页journal of Computer Applications

摘　　要：相位相关法广泛用于图像配准中,结合傅里叶梅林变换的相位相关算法可以精确检测出图像平移,旋转和缩放参数。针对传统方法,设计了一种边缘模糊滤波器来抑制相位相关法的边缘效应,同时定义了一个简化角度模糊判断的shift操作。为解决因多次应用傅里叶变换带来的计算量大、实时性差的问题,提出并实现了在GPU并行处理下的解决方案,并将其应用在实时性要求较高的工业图像配准和复杂背景下的模板定位中。实验结果证明该方法的有效性和较好的鲁棒性,也证明了处理较大尺寸图像时具有较好的实时性。Phase correlation method combining with the Fourier-Mellin transform, is a widespread registration algorithm. According to the traditional method, an edge-inhibition fillter was designed to cut down the edge effect of phase correlation. In addition, a shift operation was designed to simplify the angle ambiguity. Efficient implementation of the proposed approach based on GPU parallel process was presented to solve the problem of multiple calls of time-consuming Fourier transform. Hence, this approach is more applicable to real-time industrial images and pattern localization with complex scenes. Experimental result verifies the effectiveness and robustness of this approach, and real-time performance for large size images.

关 键 词：相位相关 傅里叶梅林变换 图像配准 模板定位 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]