Image Denoising Algorithm Based on Dyadic Contourlet Transform
This paper constructs a dyadic non-subsampled Contourlet transform for denoising on the image, the transformation has more directional subband, using the non-subsampled filter group for decompositing of direction, so has the translation invariance, eliminated image distortion from Contourlet transform’s lack of translation invariance. Non-subsampled filter reduces noise interference and data redundancy. Using the feature of NSCT translation invariance, multiresolution, multi-direction, and can according to the energy of NSCT in all directions and in all scale, adaptive denoising threshold. Experimental results show that compared to wavelet denoising and traditional Contourlet denoising, the method achieves a higher PSNR value, while preserving image edge details, can effectively reduce the Gibbs distortion, improve visual images.
Zhu Zhilin & Patrick Wang
F. Hui et al., "Image Denoising Algorithm Based on Dyadic Contourlet Transform", Applied Mechanics and Materials, Vols. 40-41, pp. 591-597, 2011