The Study On The Algorithm Of Visible And Infrared Image Fusion Based On The Quality Of Source Images

Image fusion is the process of obtaining a comprehensive image, in whichcomplementary or redundant information of the different sensors’ images is processed. Imagefusion can improve the clarity and identifiability of the source images, and enrich thedescription of the scene. Image fusion is widely used in military, medicine, geology, weatherforecast, artificial intelligence and other fields. Visible image has rich texture information andhigh resolution, but it is constrained by the light conditions. While infrared image has highanti-interference ability, especially in the night and bad weather, but the resolution is low andthe edges are fuzzy. The visible images and infrared images’ features are distinct, but at thesame time, they have strong complementarity. Therefore the study of visible image andinfrared image fusion has important practical significance, and it is also a hot spot in the fieldof image fusion. The main work includes:(1) Image registration is the premise of image fusion. As for pixel registration, and theregistration process does not have mark of the features, image registration algorithm based onHarris corner detection is put forward. Firstly, two test images’ feature points are obtainedusing the Harris corner detection algorithm. With the corresponding feature points coordinates,the transformation relation between the two test images is solved, after resampling,interpolation and denoising of the experimental image, the experimental registration image iscompared and analyzed with the reference image. The experimental results show that theexperimental images’ registration effect is good.(2) The current problems in image fusion algorithm are analyzed, and the algorithm ofvisible and infrared image considering the quality of source images is put forward. Based onthe theory of image fusion, the image quality assessment method is introduced, and the basic framework of image fusion considering the quality of source images is constructed.(3) The comentropy in image local weight is as the visible and infrared source imagesquality assessment index, the assessment results are as the weights coefficients of wavelettransform for image fusion, and the average gray, standard deviation, entropy, spatialfrequency are as the indexes of the fused image quality assessment. The algorithm of thevisible and infrared image fusion based on the comentropy in image local weight is proposed.In the end, several groups of visible and infrared images are experimented. The algorithm ofvisible and infrared image fusion considering the quality of source images is completelyfeasible and available.