随着传感器技术的发展, 单一的图像传感器往往不能够从场景中提取足够多的信息, 需进行多源图像融合。为了解决多传感器图像所表现的目标特征不一致的问题, 本文采用小波变换对红外及可见光图像进行了融合。首先利用小波变换将图像进行多尺度分解。对于高频部分融合, 取两幅图像小波系数矩阵对应元素的最大绝对值构造小波系数矩阵; 针对低频部分融合, 采用基于领域像素相关和基于区域方差相结合的策略。实验结果表明, 该算法将红外与可见光图像对同一目标所表现出的不同特征、细节有效地融合在一幅图像里, 增加了单幅图像的信息量, 丰富了目标的信息层次, 为图像显示观察和后续图像处理系统获取信息提供了基础。

With the development of sensor technologies, a single image sensor can not extract enough information from a scene. So, the multi-source image fusion is urgent to be studied. To solve the problem that the features of a same object in the images from multiple sensors are not identical, the wavelet transform is used to fuse the infrared and visible images. Firstly, the wavelet transform is used to decompose the image in multi scale. For high frequency fusion, the coefficients with maximum absolute values in two original images are selected. For low frequency fusion, the strategy of combining the domain pixel correlation with the regional variance is used. The experimental result shows that this fusion algorithm can fuse the different features and details of a same object in infrared and visible images into one image. It can increase the information in one image and enrich the information hierarchy of the target. It provides the basis for the image display and subsequent image processing.