Natural-color-appearance Steerable Pyramid Color Fusion for Visible and IR images

Based on the multi-scale and multi-orientation characteristics of Steerable Pyramid, a novel color fusion method of visible and IR images for observation is proposed. After Steerable Pyramid decomposition, the fusion method is realized in YUV color space according to the human perception of luminance and chrominance. In luminance channel Y, the local energy and matching measure are utilized to reserve the salient information and complementary information from visible and IR images. In chrominance channel U and V, linear combinations of sub-bands are performed. Then the color fused image with near natural color appearance is obtained. It can also help observers to detect and recognize the hot targets because the targets appear in orange in the fused image. The objective fusion evaluation and color statistics shows that the proposed method could reserve the detail information of visible and IR images, and greatly improve the color appearance and target detection ability.