The Image Fusion Based On Gradient Information In Contourlet Transform Domain

Image fusion aims to capture the image data of the same target from multi-source channels and use their existing information to complement each other to obtain high quality image through image processing technology.Image fusion can greatly improve the utilization of image information,enhance the visual effect of the image,improve the resolution of the original image,and improve the accuracy and reliability of computer-interpreted images.Therefore,it is widely used in medical diagnosis,geographic remote sensing,weather forecasting and navigation guidance.The research of image fusion has always been an important issue in the field of image processing.Image fusion methods based on transform domain have attracted the attention of the majority of researchers,which have the advantage of no block artifacts and rich image details.What's more,this direction is very hot now.Based on systematically studying the theory of multi-scale geometric analysis,this paper focuses on the in-depth study of image fusion based on Non-Subsampled Contourlet Transform(NSCT).In particular,the effectiveness of using image NSCT domain gradient information to improve the quality of fused images is discussed.The main work and contributions of the dissertation include:(1)An image fusion method combining NSCT with morphological gradient was studied.Given the excellent multi-scale multi-directional and shift-invariant of NSCT,a morphological gradient information embedded image fusion scheme in NSTC domain(NSCT-MG)is proposed to solve the shortcoming caused by fusing the low-frequency components of the image through the average method.To be specific,firstly,multi-source images are decomposed by Non-Subsampled Contourlet Transform.Then,Multi-scale morphological gradients are used to compare pixel information at the same position in multi-source low frequency components to generate corresponding information decision map.Finally,based on the information decision map,the low-frequency coefficients of the source image,which are useful for information,are selected to realize the coefficients fusion.Because the gradient information better reflects the details of the edge of the image,this method can effectively enhance the contrast of the fused image,especially for fusion of infrared and visible light images.(2)An image fusion method based on weighted gradient in NSCT domain(NSCT-WG)is proposed.The basic idea is to use the saliency information for fusion rule design,which is extracted from the low-frequency components decomposed by NSCT.To achieve this goal,we propose a multi-scale gradient weighted image fusion scheme in the NSTC domain,for the reason that the saliency information such as edges and corners in the image can be obtained by the gradient domain method.First,the large scale weighted structure tensor is used to extract the significant information of the low-frequency components used as the fusion coefficients.At the same time,the edge region is set.Then the small scale weighted structure tensor is used to determine the weight of coefficients in edge region around salient information.Through these two steps,the saliency information in the source image is preserved while the spatial discontinuity existing in the fused image is reduced.So,the fused images are better for visual fusion effects and observability with higher contrast.