Multi-focus image fusion method

Abstract

The invention relates to a multi-focus image fusion method and belongs to the field of image processing. An enhanced differential evolution method and an image fusion method of an expansion block selection mechanism are adopted in the multi-focus image fusion process. Compared with a standard differential evolution method, the enhanced differential evolution method is used for judging population after iteration every time according to the sizes of images and interval initial population; one individual is reserved on the premise that all the individuals in the population are the same; the restindividuals are reinitialized and served as a new generation of population together in the corresponding interval by an initialization method of a standard differential evolution algorithm, and localsearch is enhanced, so that the optimal block size of the image is acquired; in addition, when the fitness of image blocks is equal, extension blocks are selected, and the fused image blocks are acquired by comparing. According to the multi-focus image fusion method, the efficiency of searching the optimal block size in the process of the image fusion is improved, and the computational complexityis reduced; and in addition, the clarity of the whole fused image is fully considered in the process of the image fusion.

Description

technical field [0001] The invention relates to a multi-focus image fusion method, in particular to a multi-focus image fusion method based on an enhanced differential evolution method and an expansion block selection mechanism. Background technique [0002] The block-based multi-focus image fusion method makes full use of the characteristics of multi-focus images, divides the source image into several sub-blocks, and puts the clear sub-blocks in the source image into fusion by judging the characteristics of the corresponding sub-blocks in the source image. On the corresponding position of the image, the fusion image is obtained by reconstruction. The key of this method is to determine the size of the optimal block and to process the corresponding image block with equal definition. [0003] Using the standard difference evolution algorithm to find the optimal block size to realize the fusion of multi-focus images is one of the methods with simple calculation and good fusion ...

AI Technical Summary

Problems solved by technology

However, the performance of the standard difference evolution algorithm is low. As the iteration progresses, it is easy to fall into local convergence, and the population is randomly initialized, and the global search ability is low.

In recent years, many improved differential evolution algorithms have emerged, generally improving the robustness of the algorithm, convergence speed, local search and global search capabilities, such as algorithms based on chaotic local search, changing mutation strategies, and adaptive parameters. However, the above methods have problems such as cumbersome algorithm implementation, complex calculation, and low efficiency to varying degrees.

[0004] In the process of image fusion, the processing of corresponding image blocks with equal sharpness (that is, fitness) mainly has the following methods. For the corresponding blocks with the same sharpness, F i =(A i +B i ) / 2 to choose, but the experiment proves that A i and B i The corresponding pixel values ​​are not necessarily completely equal. The fused image will change the pixel value in the original image and enhance the block effect. The second is to randomly select a piece. This method is also unreasonable.

Images