Research On Super-resolution Reconstruction Of Remote Sensing Image Based On Haar Wavelet

Super-resolution reconstruction technology of image is used to enhance the spatial resolution of remote sensing image more and more extensively,and it provides more possibilities for improving remote sensing image quality.With the advantage of wavelet decomposition,the super-resolution reconstruction technology based on wavelet reconstruction can recombine components to obtain higher resolution images.Combined with the related knowledge of statistics and neural network,the reconstruction effect can be improved.This thesis use single remote sensing image block as the data basis,wavelet transform theory as the theoretical basis,combined with Hidden Markov Tree(HMT)model,interpolation and BP neural network learning methods to explore the super-resolution reconstruction technology.The main contents and conclusions of the study include:(1)This thesis based on wavelet domain Hidden Markov Tree(HMT)model.This study emphasizes the similarity between Markov Tree model and wavelet reconstruction.The super-resolution problem of image is regarded as a constrained optimization problem to reconstruct the optimal solution by iterative constraints.(2).This thesis analyzes the algorithm of super-resolution reconstruction which combines wavelet transform and interpolation algorithm.The experimental results show that the proposed method is more effective than the interpolation algorithm,and by processing and comparing with the high resolution image,the spatial resolution image can be improved obviously,and the reconstruction speed is faster.(3).This thesis focuses on the super-resolution reconstruction algorithm which has combined wavelet transform and BP neural network.Based on the similar autocorrelation between wavelet transform components and the advantage of BP neural network learning mapping relationship,this thesis has constructed a self-learning reconstruction algorithm by combining the advantages of the two algorithms.Finally,wavelet reconstruction theory is used to complete the reconstruction.According to the theoretical research,related algorithm improvement and experimental analysis above,the results of this thesis have certain reference value of reconstructing the super-resolution of remote sensing image based on wavelet reconstruction for the further research.