Single-band infrared image colorization algorithm based on color transfer

Single-band infrared image colorization based on color transfer is studied, and a method of popping-out hot targets by applying pseudo-color coding is put forward. An appropriate reference image is selected, and its luminance channel is linearly transformed. Color of the image is transferred to single-band infrared image based on decorrelated lαβ color space, through matching pixels between images using the mean and standard deviation in luminance channel over specified neighbor of pixels. The brightest parts of the resulted image are highlighted using the red hue of rainbow color coding, which usually associates with high temperature. By doing those to kinds of single-band infrared images, images with natural color appearance are obtained, which can avoid visual fatigue after long-time observation. And as the hot targets in the scene are made more noticeable, the target identification ability can be efficiently increased while the scene depth perception is remained.