Remote Sensing Images Upsampling Based on Projection onto Convex Sets and Complex Wavelet Packet Transfer
Zhang Yan · Chinese Journal of Computers · 2011
Qualities of remote sensing images are always degraded by many factors,such as atmosphere,optical blur,motion,sensors and down-sampling and so on.The authors adopted the Projection onto Convex Sets(POCS) approach,which is usually used in multi-frames images super-resolution,for an initial up-sampling of single remote sensing image.To further improve the results,the authors used Bayesian method to adaptively estimate the modulation transfer functions(MTF),and finally refined the images in Complex Wavelet Packet Transfer(CWPT) domain to reduce remaining blurs and noises.Experiment results demonstrate that the method based on POCS and CWPT performs well for remote sensing images up-sampling in a rapid and effective way,particularly for directional textures compared to traditional methods.