Research On Remote Sensing Image Fusion Algorithm Based On NSCT Transform

Image fusion as a modern high technology combines a series of technologies,such as the sensor,image processing and computer.The remote sensing image fusion,as an important branch,has drawn more and more attention.In recent years,MS images and PAN images fusion technology plays an important role in both military and civilian field.MS images have high spectral resolution and PAN images have high spatial resolution.However,in practical application,people always hope to get these images that have both high spatial and high spectral resolution,so the MS and PAN images fusion technology and the algorithm just arise at the historic moment.But some researches has demonstrated that most of the remote sensing image fusion algorithms are a pair of mutual suppression contradictions in keeping the spectral information and enhancing the spatial details.Therefore,how to achieve the goal of maintaining the spectral characteristics and simultaneously improving spatial information is still a research focus in the remote sensing image fusion.In this paper,the remote sensing image fusion methods based on multi-scale geometric transform are studied.The main ideas and implementation of multi-scale geometric transform are introduced,on this basis,the research mainly focuses on the following four aspects:1.Based on the advantages that the PCA transform doing better in improving the image spatial resolution and the Curvelet transform being beneficial to retain the image edge,we propose a new remote sensing image fusion method on the basis of PCA and Curvelet transform.2.Combined with the adaptive weighted fusion rule based on local energy,we propose a new remote sensing image fusion method on the basis of IHS and NSCT transform.3.For the sparse matrix of the RPCA decomposition can extract the salient information of the PAN images,and global coupling and impulsive synchronization of pulse coupled neural network(PCNN),so a remote sensing image fusion algorithm based on robust principal component analysis(RPCA)and adaptive pulse coupled neural network(PCNN)is proposed.4.For the remote sensing image fusion algorithm based on HSV and non-subsampled Contourlet transform has high calculation complexity and can not extract details from source images effectively.Combined with the compressive sensing and multi-feature fusion rule,a new remote sensing image fusion algorithm based on NSCT transform with compressive sensing(CS)domain is proposed.The results of a large amount of experiments show that the four kinds of algorithms,which proposed in this paper are better than some existing remote sensing image fusion algorithms in subjective and objective evaluation indexes,thus the algorithms have certain engineering application value.