Future directions in a basic recursive algorithm for high-resolution imaging from multiframes

Abstract

Summary form only given. The authors discuss the use of Kalman filtering theory for tackling the problem of high resolution image reconstruction from undersampled discrete frames, where each frame is shifted with respect to a reference frame, possibly degraded by a blur and also corrupted by additive noise. The results of computer simulations for the case where the input frames are blurred by identical six-pixel motion blurs are presented. Input images have 32×32 pixels and are generated by undersampling the blurred image. The reconstructed images have 64×64 pixels, and the various cycles of reconstruction are shown. Computer simulations for the case where the (independent) frames are blurred by independent motion blurs are considered. Conclusions are drawn with respect to the speed of convergence that can be inferred from the results of simulation