Research On Visible/infrared Aerial Imaging Payload Image Fusion Algorithm Based On Multiscale Decomposition

Aerial photoelectronic imaging load is one of the main optoelectronic devices for acquiring ground image information in the air.In aerial imaging tasks,visible light detectors or infrared detectors are usually used to acquire target images.However,it is sometimes difficult to meet the target detection and recognition requirements by relying only on a single spectral segment image.In order to getting the images with more information and more obvious features,image fusion technology needs to be introduced.Infrared and visible image fusion is one of the important branches of image fusion technology,it makes full use of the complementarity of infrared and visible image fusion,fusion image can comprehensively express scene information.The paper analyzes the theoretical basis of infrared and visible image fusion,and systematically studies the main links of the image fusion system.The research focuses on how to effectively improve the quality of low-contrast visible image and how to make full use of the complementary information of the two types of images for image fusion.Aiming at the problems encountered in the practical engineering application of aerial image fusion,two algorithms the fusion system software are designed as follows:1.In order to solve the problem that the imaging quality of airborne imaging equipment is easily affected by weather and illumination,which leads to low contrast and loss of detail information of visible image obtained by the equipment in low light environment,a visible image enhancement algorithm based on layering idea is proposed.Firstly,the hybrid framework based on rolling guided filtering and Gaussian filtering is applied in the double-scale decomposition framework of visible images.Rolling guided filtering has good scale perception and edge preservation characteristics,and can have better extraction effect on high-frequency details.It can separate the high frequency detail information from the low frequency dynamic range feature edge information with the lowest number of iterations.Finally,an adaptive dynamic range compression algorithm based on logarithmic global compression method is used to enhance the visible image.Experimental results show that the proposed visible light image enhancement method can effectively improve the contrast of low light region and recover some lost details.2.Aiming at the problems that common multi-scale decomposition methods are difficult to distinguish edge,texture and other information into different levels,and no corresponding fusion rules are selected according to the features of each level,a fusion method of infrared and visible image based on multi-scale decomposition of rolling guided filter is proposed.Firstly,in the image decomposition part,a hybrid multi-scale decomposition based on rolling guided filtering and Gaussian filtering is proposed.This decomposition framework can fully extract the structural information of the source image and lay a solid foundation for the subsequent image fusion.After filtering by the decomposition framework,the feature information of each level differs greatly.In this paper,targeted image fusion rules are adopted for different levels after decomposition.The final fusion results can better retain the detailed texture information and infrared target information of the source image,avoid artifacts at the edge of the target,and are more in line with human visual perception.The proposed algorithm is compared with the classical algorithm.The experimental results show that the proposed algorithm is significantly superior to other comparison algorithms in the edge intensity index,which proves the superiority of the algorithm in edge feature extraction.The effectiveness of the proposed algorithm is verified by real aerial images,and the main and objective evaluation of the proposed algorithm is superior to the comparison algorithm.3.Aiming at the requirement of infrared and visible aerial image fusion in practical engineering,a infrared and visible aerial image fusion system software is designed and developed.The fusion system software includes three modules: image fusion preprocessing,image fusion system and objective evaluation index.Users can complete infrared and visible aerial image fusion according to their own needs,and can save the required fusion results and objective evaluation indicators.In this paper,a new visible image enhancement algorithm and a infrared and visible image fusion algorithm based on multi-scale decomposition are proposed to focus on the theoretical research of infrared and visible image fusion of aerial imaging loads and the problems faced in practical scenes.An image fusion system software for aerial images is designed and developed.It has enriched the theoretical system of infrared and visible image fusion system and has great significance to the development of infrared and visible image fusion technology in the field of image fusion.