基于红外面阵扫描图像的配准和融合

Image registration and fusion based on infrared array scanning detector

针对红外面阵扫描图像的连续拼接,提出了一种基于灰度特征的配准和融合算法。该算法首先利用Harris算子对待拼接图像的特定区域进行角点检测,并提出了一种自适应阈值方法,用于提取特征明显的兴趣点,同时限定兴趣点的个数;其次选择合适的窗口大小,利用互相关归一化(NCC)函数,对拼接图像兴趣点邻域灰度范围进行图像匹配,获得图像变换关系;最后提出了一种基于双线性变换和渐入渐出相结合的图像融合算法。结果表明,自适应阈值角点检测后的兴趣点特征明显,分布均匀且数目被阈值很好地约束,图像配准精度高,采用本文融合算法后的拼接图像,重叠区域过度平缓,不存在拼接缝和灰度跳变现象。

An image registration and fusion algorithm based on gray feature was proposed for sequence infrared array scanning images.Firstly,the corner points were detected by Harris algorithm in the selected region,and an adaptive threshold algorithm was proposed to extract interesting points with obvious features and limit the number of the corner points;secondly,an appropriate window was selected,and then the images were matched by the normalized cross cor-relation （NCC）algorithm in neighbourhood gray range of interesting points and the transform relation between them was obtained;finally,an image fusion algorithm based on bilinear interpolation algorithm and gradated in-and-out a-malgamation algorithm was proposed.The results show that the interesting points after the corner detection with adap-tive threshold have obvious features and uniform distribution,and at the same time,the numbers of the corner points are well limited.The images using the algorithm have accurate matching ability,and the final fusion image has smooth transition without evident seams and gray jump phenomenon.