Image Denoising by Curvelet Transform Based Adaptive Gaussian Notch Filter
R. Praveena, S. Mary Cynthia, S. Jacily Jemila, Tummala Ranga Babu · Advances in engineering research/Advances in Engineering Research · 2024
In this work a new adaptive Gaussian notch filter (AGNF) with curvelet transform is proposed for removing noises from Magnetic Resonance Images.MRI images corrupted by periodic noises which are occurred because of interferences during image capturing.In general, the interferences occur due to electric or magnetic circuits.These periodic noises can be identified by repetitive patterns formed in the image.Since periodic noises affects the image quality the elimination of this noise is very important.The Adaptive Gaussian Notch filter identifies noisy peak areas and eliminates corrupted regions, also size of window is varied based on size of the noisy frequencies of the noise affected frequency domain image.This window size is varied from smaller size to the size of the noisy peak areas.The Curvelet transform having very high degree of directionality and anisotropy compared with Fourier transform and wavelet transform.In which both Fourier transform and curvelet transform were used to isolate noise regions.Finally, the calculated PSNR values were compared and the best one were identified.