Color Transfer Between Grayscale Images

More than 80% information that people’s perception of the outside come from the visualinformation. Through the study of human visual characteristics, we find the human can morequickly and accurately obtain information from the color image than the grayscale image. Inthe field of film, military and medical, sometimes people can not get the color image of thescene they desired, so the hot issue needs to solve is how to improve the recognition rate ofthe human eye through grayscale image color processing by the color transfer technology.In this paper we proposed a color transfer algorithm for colorizing the visible grayscaleimages based on morphology transformation and FFCM cluster. We used morphologytransformation to process the target image first for removing the non-uniform luminanceregion. Through the initial cluster and iterative threshold segmentation, we get the targetimage which cluster accurately. Then established the correspondence between the swatches ofsource and target images to complete the color transfer of swatches which as the reference tocomplete the target image’s colorization.The infrared image and visual image have the different imaging mechanism, so theinfrared image can’t be processing well through the algorithm for colorizing the visiblegrayscale images. In our paper we proposed a new color transfer algorithm for colorizing theinfrared images based on contrast enhancement. At first, we deal with multiple images to getthe initial pseudo-color fusion’s image which is the target image of color transfer algorithm.We introduce a contrast enhancement factor to enhance the chrominance channels of thetarget image when we select the appropriate reference image to transfer the global color for the target image. Our algorithm is applicable not only to multi-band image, but also applies tosub-band image.