Title :
Wavelet Thresholding Techniques in MRI Domain
Author :
Prinosil, Jiri ; Smekal, Zdenek ; Bartusek, Karel
Author_Institution :
Fac. of Electr. Eng., Brno Univ. of Technol., Brno, Czech Republic
Abstract :
This paper deals with MR image de-noising by using the wavelet analysis focusing on the wavelet thresholding techniques and the threshold estimation. Hard, soft, semi-soft and non-negative garrote thresholding techniques are described and applied to test images with two different threshold estimators; one uses the universal threshold and the second is derived from the Bayesian risk minimization. The results are compared according to three parameters: SNR, intensity contrast and intensity gradient.
Keywords :
belief networks; biomedical MRI; estimation theory; image segmentation; medical image processing; wavelet transforms; Bayesian risk minimization; MRI; hard garrote thresholding; image de-noising; nonnegative garrote thresholding; semi-soft garrote thresholding; soft garrote thresholding; threshold estimation; wavelet analysis; wavelet thresholding techniques; Image analysis; Image denoising; Linear approximation; Magnetic resonance imaging; Noise reduction; Signal to noise ratio; Testing; Time measurement; Wavelet analysis; Wavelet domain; MR images; threshold estimation; thresholding techniques; wavelet analysis;
Conference_Titel :
Biosciences (BIOSCIENCESWORLD), 2010 International Conference on
Conference_Location :
Cancun
Print_ISBN :
978-1-4244-5929-2
DOI :
10.1109/BioSciencesWorld.2010.14
