Super Resolution Reconstruction Of Water Body Remote Sensing Image Based On Anisotropic MRF

Remote sensing technology has become the main means of water resources and water environment investigation,water information extraction,and from remote sensing images to obtain this information,the image resolution is essential.High-resolution images can provide more accurate information to the application of remote sensing image,such as water information extraction.Super-resolution reconstruction technology gives us a low-cost method to get high-resolution remote sensing images by using the low-resolution remote sensing images.But traditional algorithm of super-resolution reconstruction is based on the isotropic,and it does not take into account the geometric structure of pixel information,and reconstruction in full consideration of the anisotropy of the images is conducive to better maintain the edge and geometry structure of images.In this paper,I have studied some super-resolution algorithms based on the MRF model and the comparison of different MRF models algorithm.The traditional MRF super-resolution reconstruction algorithm does not consider the geometric structure of the pixel,in order to make full use of the structural properties of pixels in reconstruction,keep the images structure better,we need to consider the pixels of anisotropy in apply MRF model to smooth the local window.Structure tensor is a good method to estimate the direction and properties of local geometric structure,and the bilateral filter can effectively integrated from the geometric distance and gray similarity,which can effectively restrain the noise and keep the edge of image in the image smooth.After studied the bilateral filter,the generalized MRF model and the structure tensor,I combined geometric distance measure of bilateral filter with Edge-adaptive Structure Tensor to build a anisotropic neighborhood matrix,so I can reset the weighting of the pixels in neighboring window by its geometric distance and direction information.Finally,Combined with the adaptive threshold Huber function,a sensing image super-resolution reconstruction model was built based on anisotropic MRF.Through the experimental and compared with the original reconstruction algorithms based on MRF model,the new model can keep the geometric information of the image,especially the edge information,and can effectively improve the accuracy of river extraction.At the end of this paper,I summarized the research and the achievements of this paper,making a further prospect for the future work and puts forward some suggestions.