Restoration of Blurred and Noisy Images Using Inverse Filtering and Adaptive Threshold Method

Abstract

A restoration scheme for images that are corrupted with both blur and impulsive noise is proposed in this paper to reconstruct an image with minimum degradation. The restoration scheme consists of two stages in sequence where the first stage is applied to the blurred image and the second stage is applied to de-blurred image that has been subject to noise through electronic transmission. The first stage uses frequency domain filtering while the second utilizes spatial filtering to reduce the indicated blur and noise, respectively. In particular, truncated inverse filtering is used for reducing the blur and an adaptive algorithm with an estimated threshold is used for minimizing the noise. Simulation of the introduced method uses several performance measuring indices such as mean absolute error (MAE) and peak signal-to-noise ratio (PSNR). Results of these simulations show great performance of the proposed method in terms of reducing the blur and noise significantly while keeping details and sharpness of the image edges.