Optimization Of Image Super-resolution Reconstruction Algorithm Based On Fractal Features

The real-time nature of remote sensing images is more conducive to detecting natural disasters in some ways.However,due to the influence of equipment and other conditions,the resolution of single frame remote sensing images often decreases.Image super-resolution reconstruction technology can greatly improve the resolution of remote sensing images.This thesis mainly studies the super-resolution reconstruction algorithm of remote sensing image,classifies the image blocks in training samples by using the idea of local fractal dimension in fractal theory,extracts template features by using first-order and second-order features respectively,which makes training samples more refined;optimizes K-SVD dictionary learning method and uses OMP algorithm to sparsely encode atoms without repetition,so as to make the iterative process.The error is smaller and the dictionary information obtained is more abundant.By combining the two points and using PCA to reduce dimension,the efficiency of the whole algorithm is improved.The simulation results show that the numerical values of PSNR and SSIM are better than those before optimization,and the quality of image reconstruction is improved.