Study On Multi-focus Image Fusion Algorithm Based On Block Mechanism

With the development of image sensor, the application of image fusion technologyhas become increasingly widespread. It is difficult to make all the objects focused in thesame scene. So how to fuse the objects in order to make all the objects clearly in thefused image has become imminent.This thesis focuses on the multi-focus image fusion algorithm which is based onblock mechanism：①A multi-focus images fusion method which is based on enhanced differentialevolution algorithm and extends block selection mechanism is proposed. Differentialevolution algorithm-based multi-focus images fusion has been demonstrated to be asimple, efficient and robust method. However, there are still some problems. Firstly, itmay easily fall into the local convergence. Also, it may serious impact on the globalsearch ability. In order to solve these problems, an enhanced differential evolutionalgorithm is presented in this paper. Secondly, when the sharpness values are equal ofthe corresponding blocks, which may change the values of the source images andenhance the block effect. This paper presents a simple and effectiveapproach——extend block mechanism, which is not change the value of the originalimage and reduce the block effect. The modified differential evolution algorithm andextend block to select the same sharper blocks can get the better fusion results with alarge number of experiments in this paper.②In order to improve the efficiency of images fusion, a new multi-focus imagesfusion algorithm which is based on enhanced differential evolution algorithm andadaptive block mechanism is proposed. Firstly, fusion of multi-focus images usingenhanced differential evolution algorithm and extends block selection mechanismignores the factor of image size, which handle all pictures with the same way, thenlarger size pictures will calculate complexity and take a long time. Secondly, themethod do not take full advantage of multi-focus image characteristics，which is dividedinto a clear, fuzzy and the boundary region. All clear area and fuzzy area still continueto divide, which cause unnecessary calculations and time spending. Block algorithmusing adaptive multi-focus image fusion only continue to divide all the boundary region.However，the boundary region become smaller and smaller to further reach the exactlydividing point，which is relatively complicated and not precise enough to select the best block size and the threshold parameter. A multi-focus image fusion method usingenhanced differential evolution algorithm and adaptive block mechanism is proposed inthis paper, which combine the advantages of the above two methods and make up foreach other’s shortcomings. Experimental results show that this algorithm not only canget better fusion effect, but also can calculate more simply and spend less time.