Multi-focus Image Fusion Based On Random Walk And Guided Filter

In recent years,random walk and guided filter have got more and more attention in image proposing,with the former used in image fusion,image smoothing,image enhancement,image segmentation and so on.The latter has good advantages of smoothing and edge-preserving,so it is popular with researchers.We propose two multi-focus image fusion methods and the main idea of our methods is applying both the random walk and the guided filter to image fusion,which can get their advantages together.This paper mainly propose two multi-focus image fusion methods based random walk and guided filter.The first algorithm first uses random walk filter to decompose the source image into two scales;then uses random walk filter and guide filter to obtain the initial weights of detail layer image and basic layer image respectively;then combines the corresponding two initial weights by different proportions to obtain the final weights of the basic layer and the final weights of detail layer;finally,the fused image is obtained by weighted averaging and summation steps.The second algorithm uses random walk filter to decompose the source image into two scales.Then the guided filter is used to optimize the initial weights,and the intermediate weights are obtained.Then the random walk filter is used to optimize the intermediate weights,and the final weights are obtained.Finally,the fusion image is obtained by weighted averaging and summation steps.