基于融合图像特征库的夜视图像彩色化研究

Research on night vision image colorization based on the fusion of image′s character library

提出一种基于图像融合技术的夜视图像彩色化方法,结合红外图像和微光图像各自的特点,利用小波图像融合的方法把红外图像和微光图像进行融合,从大量的融合图像中提取不同物体的纹理特征,构成野外环境下的纹理图像库,并对每类图像指定相应的彩色参考图像。对目标图像进行分割,提取目标图像的纹理特征与图像库进行相似性对比,完成目标图像的色彩传递。实验结果表明,该算法获得真实的场景色彩,使观察者更容易获得图像的场景信息。

This paper describes a Night Vision Image Colorization based on image fusion technology. Concerning their characteristics,low-level visible and IR images are combined into a fusion image using wavelet fusion. We extract dif ferent patterns of textures to construct a library based on lots of experiments, and the color images are destined for dif ferent patterns. We segment the target image and compute the texture of different patterns. And the color transformation is achieved. The experimental results show that the algorithm can make a color night vision image more realistic and natural, helps observers to get more abundant scene information.