Fusion Of Infrared And Low-Level Light Images In Realistic Scene

In recent years,with the rapid development of science and technology,more and more multimedia devices have become an indispensable part of people's lives..In order to meet the increasing demand for image quality,image sensor technology and various digital image processing methods are getting more and more attention.However,due to the impact of its imaging performance,single sensor is difficult to fully reflect all the effective information in the scene,so it is difficult to meet the actual needs in certain scenarios.In order to synthesize the imaging characteristics of various sensors and avoid redundant information between different types of images in a specific scene,and to provide a more comprehensive and effective information description of the scene,image fusion algorithm has been proposed and attracted more and more attention in recent years.Especially infrared and low-light-level image fusion technology,as an important branch of multi-spectral image fusion technology,can be used in many fields such as security monitoring,night combat,night driving and so on.Under the above research background,this paper uses the advanced low-light-level infrared acquisition equipment to get the infrared low-light-level image pairs in complex realistic scenes.In order to avoid the imaging defects of the previous low-light-level and infrared image algorithms in complex scenes,a new image fusion method based on guided filter is proposed in this paper.The main contents of this paper are as follows:Firstly,this paper briefly introduces the basic theory of low-light-level infrared image fusion and the mainstream image fusion algorithm,analyzes the defects of current low-light-level and infrared data sets,and uses the current advanced image acquisition equipment to construct image pairs in real scenes,and finds out that the current low-light-level and infrared data sets have defects.Image fusion is a defect of image fusion effect for long distance,multi-target and strong light interference.Aiming at the problem that the remote characters are not obvious in the fused image,an infrared target extraction algorithm combining spectral residuals and GBVS is proposed.Firstly,multiple infrared targets in the fused image are detected and extracted.Finally,the fused image is enhanced by a certain weight,so that the algorithm can be effectively implemented.Infrared target will not be lost in remote detection,so the characteristics of infrared remote imaging fully reflect in the fused image.Then,combining with the high-resolution imaging ability of the current advanced low-light-level sensing equipment,the corresponding contrast enhancement algorithm is proposed to enhance the contrast information in the low-light-level image,so that the detail information of the low-light-level image can be fully displayed on the fused image,and the strong light interference in the low-light-level image can be suppressed.At the same time,the strong light interference in the low light level image is suppressed to ensure the observability of the information around the strong light source in the fused image.Finally,an image fusion algorithm framework based on guided filter is proposed,which can decompose the image into a base layer and multi-scale detail layers.The base layer is fused by saliency detection algorithm and the detail layers are fused by region energy maximization method.The method can enhance the infrared target and solve the problem of low-light and infrared image fusion in special scenes such as remote multi-target detection,strong light interference.