A Block Image Fusion Algorithm Based on Algebraic Multi-grid Method

In this paper, a block image fusion algorithm based on algebraic multi-grid (AMG-Block) method is proposed. The main novelty of this idea is that the algebraic multi-grids method has good abilities to extract the structure information of images, which overcomes difficulties faced by the existing fusion methods. At first the rough grid data of the image is extracted and reconstructed. Because the mean square error between clear image and reconstructed image is bigger than the mean square error between fuzzy image and its reconstructed image, this characteristic can be used to judge the sharpness of the image. Then the sliding window tactics is used to make the effect better, and the decision map is obtained. At last the final decision map is generated and rectified with the assist of the surrounding pixels. Experimental results demonstrate that the proposed method can obtain better fusion performance on the part of both visual quality and objective evaluation.