High-quality Parameters Estimation for Noisy Motion Blurred Image Restoration

A new scheme with high performance and anti-noise capability is presented for restoring a noisy motion blurred image. The Fuzzy method based on local mean square error and mathematical morphology algorithms are utilized to deal with power spectrum of the image, and the empirical formula is offered for estimating the parameters and the method for defining the morphological template. Then the angle accurately via Radon transform can be obtained. Experiments show that motion blur angle, motion blur length and the variance of image have a nonlinear function relationship. So, the 3D surface is formed by the three factors and it fits the polynomial expression. Then, the length fast can be estimated via the expression. At last, this paper carries out simulations to validate the proposed scheme on blurred images with different motion parameters and different noise. Results show that it can estimate the motion parameters more efficiently and accurately than counterparts even in a strong noisy environment.