Research On The Fusion Algorithms Of Infrared And Visible Image

Infrared and visible image fusion is a backbone branch of the image fusion field and is widely used in many fields such as scene surveillance,driver assistance,public security and biometrics.Infrared sensors and visible light sensors are the most common and widely used sensors in everyday life,and because of their different imaging principles,the acquired images also exhibit their own different characteristics.Infrared sensors identify targets based on the amount of thermal radiation information between the target and the scene,and have the advantage of strong anti-interference ability and can work 24 hours a day,but the images obtained have the disadvantages of poor visibility,low clarity and lack of detailed information.Visible light sensors capture data from images through the reflective properties of light,and the images acquired have the advantages of sufficient spectral information,high clarity and good visual perception,but they are easily affected by factors such as weather and light levels.The fusion of infrared and visible images can effectively utilise their complementary information to increase the amount of information available for scene description,improve the spatial resolution of the images and expand the spatial range of target detection.Therefore,it is important to conduct research on the fusion of infrared and visible images.This thesis focuses on infrared and visible image fusion algorithms.Firstly,the imaging differences and unique characteristics of infrared and visible sensors are analysed and compared,and the basic theory of image fusion and the quality evaluation system of fused images are studied.Secondly,the principles and processes of four classical image fusion algorithms are introduced in detail,and the advantages and disadvantages of their algorithms are analysed.Finally,two fusion algorithms are proposed to address the current problems of low contrast of fused images,blurred target contours,easy loss of background texture detail information and inconspicuous highlighting of salient targets in the infrared and visible image fusion algorithms,mainly as follows:1.Aiming at the problems of poor contrast and blurred target contours in traditional infrared and visible image fusion algorithms,a guided filtering-based infrared and visible image fusion algorithm is proposed.Firstly,an adaptive enhancement method based on dynamic range compression and contrast recovery of guided filtering is used to improve the visibility of the dark region of the visible image;secondly,a multi-scale decomposition of cross-bilateral filtering is performed on the image to be fused to obtain the detail layer and the base layer image;a fusion method combining absolute value taking large strategy and guided filtering is used to fuse the base layer image;finally,the fused base layer images and detail layer images are summed to obtain the fused images.After experimental simulation,the fused image obtained by this paper has high contrast,clear target and outstanding details,and has better fusion accuracy and visual effect compared with other algorithms.2.In order to further improve the quality of the fused image,the Nonsubsampled Shearlet Transform(NSST)and Pulse-Coupled Neural Network(PCNN)are proposed in order to make the fused image have better visual perception for human sights.The proposed method is based on NSST and PCNN.Firstly,the NSST transform is used to decompose the infrared image and the visible image to obtain high frequency sub-bands and low frequency sub-bands containing different features of the source image.Secondly,the high-frequency sub-bands representing the details and textures of the source image are fused using a PCNN model-based fusion strategy,and the low-frequency sub-bands representing the contour information of the source image are fused using an energy attribute-based fusion strategy.Finally,the fused image is reconstructed by NSST inverse transform.Through experimental simulation,the proposed method effectively preserves the background detail information of the source image,highlights the salient target information and has a good visual effect.