Research On The Multi-focus Image Fusion Algorithm Based On Algebraic Multi-grid Method

As a branch of information fusion technology,image fusion technology has high research and application value in the field of image processing.The multi-focus image fusion technology uses an image fusion algorithm that processes multiple images from different focus areas in the same scene and uses the complementary information to generate a fused image.Compared with the source image,the fused image can describe the scene more clearly and completely.This thesis researches the existing multi-focus image fusion methods and applies the algebraic multi-grid method to image fusion.In the process of applying the algebraic multi-grid method to image processing,the reconstructed image that corresponds with the source image is generated.After experimenting,it is found that the value of the mean square error between the source image and the reconstructed image has a great relationship with the sharpness of the two images.By using this feature,the region definition of the source image is judged.The work of this thesis includes the following two aspects:1.A block image fusion algorithm based on algebraic multi-grid method is proposed.Firstly,using the mean square error of the source image and the reconstructed image,the sliding window is used to determine the sharpness of the image block that the same size as the sliding window.Choose the clear image block from the source image,and the initial decision map is generated.Secondly,the decision map is blocked,each pixel that the source is incorrect is rectified,the final decision map is generated and it would guide the image fusion.2.An image fusion algorithm based on algebraic multi-grid method and watershed segmentation is proposed.Firstly,a watershed image segmentation algorithm is used to segment an image that is one of the multi-focus source images,and several image regions with different sizes are obtained.Secondly,the mean square error of the source image and the reconstructed image of the segmented image region is calculated which the sharpness region is chose,the region sharpness decision map is generated.Finally,the region sharpness decision map is adjusted,and the final decision map is generated,the fused image is gotten.In this thesis,the relevant experiments are done to prove the superiority of the above algorithms,and the two algorithms are compared.