Research On Multi-exposure Image Fusion Algorithm

With the development of modern science and technology,people are increasingly eager for the pursuit of high-definition visual experience,which requires us to use more and more advanced camera equipment,but compared to the most intuitive visual experience of humans,cameras have certain technical defects.For example,in a particular exposure state,the imaging device is not able to capture all the details of the image in the original ecological state,and the image of a particular location of the image may still be unclear.In order to solve this series of problems,multi-exposure image fusion algorithm came into being and gradually became the mainstay of image fusion algorithm,which was explored by more researchers..Traditional multi-exposure image fusion algorithms usually get poor results of the final fusion result map,making the visual effect of the image unclear.Through the reading of a large number of documents,this paper introduces a multi-exposure image fusion algorithm,and introduces a multi-exposure fusion algorithm combined with guided filtering and a detailed enhanced multi-exposure fusion algorithm.For the problem of halo and gradient inversion caused by guided filtering,and some edge information in image fusion is lost during processing,a new adaptive multi-exposure image fusion algorithm with improved pilot filtering is proposed.Firstly,in the guide filter,the weight function is set according to the gradient information,and a function is created combining the pixel points of the image and the average value of a certain area to jointly realize the adaptation of the texture characteristics of different regions;secondly,the average brightness of the image and the contrast of the image and the saturation of the image are used.The relation between degree and the moderate degree of exposure of the image,setting the weight function,so that the weight value in the process of weighted average fusion is no longer a fixed value,and can adaptively be based on different image brightness,the weight value is also different,and finally The quality of the experimental result graph obtained is better;finally,a part of the detailed information in the original sequence diagram to be processed is added to the improved pilot filtered image to construct a texture detail layer.The experimental results weakened the halo and gradient reversal phenomenon,making the image more real,the details more clear,and the effect of image processing with a small light source is better.The final algorithm experimental result is obviously better than the effect of multi-exposure image fusion processing using multi-exposure fusion algorithm and guided filtering,and the highest quality of 2.5%,30%and 30%is obtained in the information entropy,mutual information and edge information evaluation respectively.