K11. Modified Hybrid Median filter for image denoising

A critical issue in image restoration is the problem of Gaussian noise removal while keeping the integrity of relevant image information. Clinical magnetic resonance imaging (MRI) data is normally corrupted by Rician noise from the measurement process which reduces the accuracy and reliability of any automatic analysis. The quality of ultrasound (US) imaging is degraded by the presence of signal dependant noise known as speckle. It generally tends to reduce the resolution and contrast, thereby, to degrade the diagnostic accuracy of this modality. For this reasons, denoising methods are often applied to increase the: Signal-to-Noise Ratio (SNR) and improve image quality. This paper proposes a statistical filter, which is a modified version of Hybrid Median filter for noise reduction, which computes the median of the diagonal elements and the mean of the diagonal, horizontal and vertical elements in a moving window and finally the median value of the two values will be the new pixel value. The results show that our proposed method outperforms the classical implementation of the Mean, Median and Hybrid Median filter in terms of denoising quality. Comparison with well established methods, such as Total Variation, Wavelet and Wiener filters show that the proposed filter produces better denoising results, preserving the main structures and details.