Research And Application Of Multi-focus Image Fusion Algorithm Based On Sparse Representation

In real life,in the face of complex shooting scenes,given the limited depth of field of optical imaging camera lens dilemma,all targets in the scene can only focus on different distances.The corresponding visual effect is that several objects are in focus,imaging is clear;other objects may be completely out of focus,blurred images.In order to get all-in-focus feature images,the multi-focus image fusion algorithm emerges from time to time and focuses on different objects in the scene respectively to get the source images of different targets.Then a full-fledged image is obtained through a certain fusion rule.This is the multi-focus image fusion algorithm.With the development of imaging technology,the focus of multi-focus image fusion technology is getting higher and higher all over the world,and it is applied in various fields such as machine vision,medicine,military,public safety and so on.This paper focuses on the multi-focus image fusion technology based on sparse representation.Firstly,the basic theories of sparse representation and multifocus image fusion are introduced.Then the existing technical defects of image fusion are studied in depth.Corresponding improvement schemes are put forward and the multifocus image fusion system based on double sparse representation is developed..The main research work is as follows.Aiming at the current research on multi-focus image fusion based on sparse representation,the analysis of dictionary lack of adaptability,low efficiency of learning the dictionary,and the traditional fusion strategy can not retain more image detail information,the author proposed based on double sparse dictionary Learning multi-focus image fusion method.In this method,a sparse dictionary is obtained by training a sparse K-SVD algorithm using high quality natural images as a training set.Secondly,the fused source image is segmented and drawn into a column vector using a sliding window technique,and then the sparse dictionary is calculated using a SOMP algorithm And then use a mixed poly-norms tocalculate the activity level of the image blocks.The sparse representation coefficients are fused using the rule of "selecting the largest".Finally,the fusion image is reconstructed according to the fusion coefficient and the double-sparse dictionary.Experimental results show that the proposed method has the best performance compared with the other five classical image fusion methods,and the average of the indexes and the average values ??of the two experimental groups are 7.4501 and 0.7479,respectively.By doing so,in the subjective visual and objective test All have improved.According to the characteristics of multi-focus images and multi-focus image fusion based on double sparse dictionary learning,the author designed and developed a multi-focus image fusion system software.The system is built on matlab2013 a development platform,the use of matlab language programming,the realization of seven kinds of image fusion methods in the multi-focus image fusion practice.The advantages of multifocus image fusion based on double sparse dictionary learning are highlighted by comparing the performance of each image fusion method from aspects of image information retention,image sharpness and image detail information.