基于扩散方程和相位一致的异源图像配准算法

Heterologous images registration based on diffusion equation and phase congruency

针对同一场景下的红外与可见光图像特征点难以提取和匹配的问题,提出一种基于扩散方程和相位一致模型的红外与可见光图像的配准算法。首先提出了收敛速度更快的扩散方程,并用该方程对红外图像除噪;其次利用改进的相位一致模型提取红外与可见光图像的视觉相似性结构;在相似性结构上提取特征点,进行二进制描述;最后采用汉明距离实现特征点匹配。实验结果表明,该算法能够有效滤除红外与可见光图像的差异,减小计算开销的同时实现图像的自动配准。

Aiming at the difficulty of extracting and matching the infrared and visible image feature points in the same scene,a registration algorithm of infrared and visible images is proposed based on diffusion equation and phase congruency model. Firstly,the diffusion equation with faster convergence rate is proposed,and the equation is used to denoise infrared image. Secondly,the improved phase congruency model is used to extract the visual similarity structure of infrared and visible images. Then the feature points are extracted from the visual similarity structure,and the M-LDB descriptors are used to describe them. Finally,the Hamming distance is used to realize the feature point matching. Experimental results show that the algorithm can effectively filter the difference between infrared and visible images,reduce the computational overhead and achieve the automatic registration of images.