Laplacian-Based Frequency Domain Filter for the Restoration of Digital Images Corrupted by Periodic Noise

Abstract

This paper presents a new Laplacian-based frequency domain filter for the effective restoration of images corrupted by periodic noise. In the noise detection phase of the proposed algorithm, the Laplacian high-pass filter eases the thresholding-based noise detection process by highlighting noisy peak areas from uncorrupted areas of the origin-shifted Fourier-transformed corrupted image. The filtering phase of the proposed algorithm restores the detected noisy frequencies with nearest uncorrupted frequencies to ensure high fidelity in the restored image. Experimental analysis on different images reveals that the proposed algorithm produces better outputs in terms of subjective and objective metrics than other prominent algorithms.