Region-Based Multi-focus Image Fusion Using Guided Filtering and Greedy Analysis

Region-based image fusion methods have a number of advantages over pixel-based image fusion methods. In this paper, we propose a region-based multi-focus image fusion approach using guided filtering and greedy analysis. The original images are enhanced by guided filter first and then we conduct the sparse representation of images using the greedy algorithm. Here, simultaneously orthogonal matching pursuit (SOMP) algorithm is adopted, which could obtain more accurate sparse coefficients under the same basis by processing the source image simultaneously. In order to form the regional map, the clarity enhanced image is designed and normalized cuts algorithm is adopted to segment it. According to the regional fused sparse coefficients, we recover the fused image. To verify the effectiveness of the proposed method, several pairs of multi-focus images are tested. Comparing with other fusion methods, the experiment results demonstrate that the performance of multifocus image fusion by our proposed method is superior.