基于相位一致性和Hough变换的多源图像配准方法

由于红外图像与可见光图像对比度不同，常用基于梯度幅值的特征匹配方法难以正确配准。在分析红外图像与可见光图像成像机制的基础上，提出了一种结合相位一致性边缘检测与Hough变换的多源图像配准新方法。该算法首先采用高通滤波和平台直方图均衡方法对红外图像进行预处理以提高红外图像的对比度，再利用具有图像对比度不变性的相位一致性边缘检测法提取两幅图像的边缘，结合Hough变换选取图像空间中最长的线作为特征，采用改进相位相关法作为相似性度量，在对数极坐标域下计算出两幅图像的几何变形参数。仿真实验结果表明，该方法能够以较高查准率实现红外与可见光图像自动配准，并具有较强的鲁棒性。

Multi-sensor image registration method based on phase congruency and Hough transform

Due to the contrast difference of infrared and visible images, the common registration methods based on gradient magnitude are difficult to match correctly. A novel algorithm based on the phase congruency edge detection and Hough transform was proposed by analyzing the imaging mechanism. The Gauss filter and platform histogram equalization method was applied to enhance infrared image contrast. The phase congruency algorithm with image contrast invariance was used to extract the edge of the two images. And the longest line was selected by Hough transform as the similar characteristics. Then the modified phase correlation method was used as the similarity measurement to compute geometric deformation parameters of two images under the log-polar domain. The experimental results indicated that the algorithm could achieve higher precision and robustness for infrared and visible image automatic registration.