Research On The Fusion Algorithm Of Infrared And Visible Images Based On Non-subsampled Shearlet Transform

The fusion of infrared and visible images is a research hotspot in the field of image fusion.Infrared and visible images have different characteristics.Infrared image reflects the temperature difference between the target object and the surrounding environment.Through the infrared image can overcome obstacles on the vision,even at night or harsh conditions,it can also detect clear targets,but the contrast of infrared images is low,it is difficult to reflect the details of the scene around.However,the visible image can provide clear scene details and higher spatial resolution and contrast,but the adaptability to the environment is poor.In the case of poor lighting conditions or low visibility,the image quality is poor and it can't reflect hidden hot target.Due to its good complementarity between the infrared and visible image,therefore,we can obtain a more comprehensive,reliable and abundant scenes information through the fusion of infrared and visible images to facilitate further analysis and processing.How to efficiently represent images is a core issue in image fusion,effectiveness of image representation method directly determines the quality of the fused image.Analysis shows that the wavelet is only suitable for expressing objects with isotropic singularity,for objects with anisotropic singularity,the representation of wavelet is not optimal.Contourlet transformation will cause spectrum aliasing because of the lack of translation invariance.Non-subsampled Contourlet Transform runs for a long time which is not suitable for real-time applications.Non-subsampled Shearlet Transform can overcome the above shortcomings of those multiscale transform tools,it is currently the most advanced method.Therefore the Non-subsampled Shearlet Transform is applied to the fusion of infrared and visible images in this paper.Another core issue is the fusion of image decomposition coefficients,after determining the image representation tool,the design of fusion rules determines the final fusion effect.According to the characteristics of infrared and visible images,this paper proposed a new low-frequency fusion rule based on saliency map,compared with the fusion rule based on simple weighted average or regional energy,it has obvious advantages.According to the human visual system is sensitive to image contrast,this paper presents a high-frequency fusion rules based on the improved local area contrast by analyzing the shortcomings of the conventional definition of the image contrast.Through the experimental comparison with other different fusion methods,the results show that the proposed fusion method in this paper can obtain better fusion effect,and has a certain degree of increase in objective evaluation index of entropy,standard deviation,average gradient and spatial frequency.Whether from subjective vision or objective evaluation index shows the superiority and effectiveness of the algorithm in this paper.