Multi-focus image fusion combined with CNN and algebraic multi-grid

• 1、School of Computer Science and Technology Chongqing University of Posts and Telecommunications, Chongqing 400065
• 2、School of Softwarre Engineering Chongqing University of Posts and Telecommunications, Chongqing 400065

Abstract：In this paper, the multi-focus image fusion algorithm based on CNN segmentation and algebraic multi-grid(CNN-AMG) is proposed ,for some problems such as over-segmentation problem generated by the Watershed segmentation algorithm and unstable clarity judgement by small areas. First, we use the CNN segmentation results to guide the Watershed segmentation results to merge the regions generated by the Watershed segmentation method, and then algebraic multi-grid (AMG) is used as the clarity evaluation index for each region to select the clear regions into the fusion image and perform the final fusion. The experimental results show that the fused image quality obtained by the CNN-AMG algorithm is better than the fusion image quality obtained by the traditional fusion methods with some evaluation indexes.

Keywords： Image segmentation Image fusion Algebraic multi-grid Mean square error Quality evaluation