A Study On Restoration Algorithms Of Motion Blurred Images

Image restoration is very important in the Digital Image Processing. The technique is aimed to remove motion blur, noises, and reconstruct the degraded image as possible as we can.This paper studies and analyses the theory of images blurred by uniform motion especially, the model of degradation and restoration. We propose three new improved restoration algorithms based on Hopfield Neural Network, Hopfield Neural Network and TV (Total Variation) model, and Bandelet transform:(1) An improved algorithm based on Hopfield Neural Network is proposed. We use a continuous state change network model to update the state of neurons and calculate Euclidean distance as a criterion of network stability. Experiment results show that the improved sequential algorithm could converge to a stable point and give more precise restoration results.(2) An algorithm based on improved Hopfield Neural Network and TV model is proposed. We use our improved Hopfield neural network to realize the image restoration based on TV model. Experiment results show that our algorithm is stable and has better performance and effect in image restoration superior to others.(3) We proposed a restoration algorithm based on Bandelet transform. This algorithm uses a two-step IST (Iterative Shrinkage/Thresholding) algorithm tailor for objective function of bandelet coefficients. The simulation results show the convergence and effectiveness of the proposed algorithm.