Noisy Motion-blurred Images Restoration Based on RBFN

To restore a degraded image, which has been corrupted by some kinds of noise and motion blur, a model for restoration is first presented and then an algorithm based on this model and the radial basis function network (RBFN) is proposed in this paper. In the first step of this algorithm, noise is removed by using the RBFN interpolation with variable regularization parameters. In the second step, the motion blurred image is restored based on the automatic identification of the direction and length of the motion-blur. Moreover, a method, based on object extraction, for restoring local motion blurred image is given to solve the problem of the bad restoration results when the global motion-blurred characteristics are missed for local motion-blurred images. Experimental results demonstrate that the proposed method performs well when applied to general motion-blurred images, and can also be applied to the local variable speed motion-blurred images. The restoration algorithm can give accurate identification of the degraded model and good restoration results, even under the condition of a relatively large noise interference.