Multi-Spectral Image Fusion Method Based On Sampled Three Channel Non-Separable Wavelets

Digital image fusion is the digital image processing technology that it use the sensor technology for the same target or scene of repeated imaging,according to certain rules to synthesize a new image.The result image that the comprehensive description of the scene or target is easier to be handled and perceived visually on computer can better reflect the information of the original image.The research object of the Digital image fusion is the information fusion of the image.As we all know,information fusion of multi-source information is multi-angle,multi-level processing,so as to achieve an accurate assessment of the state,identity assessment and complete and timely situation and threat assessment,to maximize the integrity of the target or scene description.With the rapid development of sensor technology and computer processing technology,digital image fusion applications are increasingly being used today and it has become a focus topic of applications including military,medical imaging,satellite imagery,computer vision,face recognition and computer vision and other fields.At present,the wavelet transform method is widely applied in image fusion,it has become one of the core image fusion research.However,since the tensor product wavelet can only capture the information of the horizontal,vertical and diagonal limited directions,it cannot fully reflect the unique characteristics of the image itself,and the use of the limited nature of the direction will cause that the image itself cannot make full use of the geometrical.The new image tensor product wavelet transform fusion method produced space regularity,therefore,generate low resolution and is easy to produce diamonds effect.In recent years,non-separable wavelets have been proposed,which solves the tensor product wavelet lack of the direction and realize isotropic,and the theories and methods of construction of the filters more general.Three Channel Non-separable is a wavelet-based scalable matrix[2,1;1,1]non-separable wavelet,which has a smaller amount of calculation of decomposition and reconstruction than the four channels non-separable wavelet,while the second channel cannot be separated wavelets,although its operation has increased,but it is due to the two-channel non-separable wavelet filter bank in addition to the diagonal,the other values are zero,while the filter will result in loss of information filtering,this paper proposes three non-separable wavelet fusion method through decomposition non-separable new framework,and applied to multi-spectral image fusion.This paper proposes a multi-spectral image fusion method based on three-channel non-separable wavelet sampling,the sampling methods and a three-channel non-separable wavelet are combined and applied to multi-spectral image fusion.Three-channel inseparable wavelet is one of the non-separable wavelets,in three-channel non-separable wavelet fusion process,the target image will decompose on every layer a low-frequency information and two high-frequency information,a lot of practice proved the feasibility and reliability of this method.In addition,this paper proposes a new sampling method,and extracts more useful information from the original image so that fusion can get more details,and enhance the performance of the new fusion image.