Open AccessA New Method for Medical Image Denoising using DTCWT and Bilateral Filter
Author(s) -
P. Vinodhbabu*,
P. Swapna
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.k1882.1081219
Subject(s) - artificial intelligence , computer science , noise reduction , thresholding , computer vision , complex wavelet transform , pattern recognition (psychology) , non local means , speckle noise , noise (video) , image quality , filter (signal processing) , wavelet , wavelet transform , image (mathematics) , discrete wavelet transform , image denoising
The quality of digital medical images plays vital role in Non-invasive imaging techniques, which are suitable for medical diagnosis and treatment. Removal of noise from a noisy image without losing the diagnostic details in medical image is still a challenging task even though several denoising methods have been proposed since past years. The wavelet thresholding approach has been reported to be a highly successful method for image denoising. However, the main problem experienced in wavelet thresholding is smoothening of edges. In order to retain original texture while denoising medical images, several methods have been reported in literature. In this paper, we proposed, a new method based on combination of dual-tree complex wavelet transform (DTCWT) and bilateral filters for denoising of medical images. The proposed models are experimented on standard medical images, like MRI image of knee contaminated with Rician noise, CT Scan image of brain contaminated with Gaussian noise, Ultrasound image of liver contaminated with Speckle noise. The results have shown that denoised images using the proposed approach have better performance in terms of smoothness and accuracy compared with existing methods. To assess quality of denoised images the quality metrics, the standard Signal to Noise Ratio (SNR), Universal Image Quality Index (UQI) Mean square error (MSR), and Structural Similarity Index (SSIM) are employed.