Polarization and spectral information fusion by an improved Welsh image color transfer method

摘要

Aiming to make polarization images more suitable for visual perception, we improve the classic Welsh color transfer method and make it more suitable for fusing underwater polarization images and spectral images. First, we segment the target image into the foreground and background regions by using super-pixel processing for the background and pixel progressing for the foreground. At the same time, the texture feature of the foreground area is introduced for feature matching. Moreover, the dark channel prior model is used to restore underwater spectral images. The given results demonstrated that the method proposed in this paper has a better performance in human visual perception than that for the classic Welsh color transfer method.