Dynamic Denoising Mirroring Filter for Periodic Noise Removal from 2D Medical Images

Abstract

The branch of image processing has been playing a critical role in various domains, including medical imaging. The challenge with medical images is the introduction of periodic noise during acquisition or transmission, degrading image quality. In medical, noisy images might affect the physician’s interpretation. We have proposed a dynamic denoising technique for 2-D medical image restoration from periodic noise to overcome this problem. In our proposed solution, we have applied the automatic detection and recovery procedure to locate the spikes to minimize the influence of degradation from the entire Fourier domain image. In this process, only the half-image band is cleaned to save time and space. A 2D Discrete Fourier Transformation (DFT) procedure is carried out to re-obtain the inferior band of the image, where deleted locations are replaced or reallocated with new bands using the mirroring or conjugate symmetry method. The algorithm is validated with visual and well-known quantitative measures on different noise strengths. The results show that, in general, our solution performs well on the highest level of noise strength.