Noise removal using double-density dual-tree complex DWT

The real world signals do not exist without noise. Image denoising system should remove this noise to recover the original signal. Noise removal can be conducted in the time-space (original signal) domain or in a transform domain. To perform in transform domain, researchers utilize the Fourier Transform (FT) or the Wavelet Transform (WT). The Wavelet Transform, specifically Discrete Wavelet Transform (DWT) performs well in noise removal applications. But they suffer from poor directional selectivity, shift sensitivity problem and absence of phase information. The proposed double-density dual-tree complex DWT is based on two scaling function and four distinct wavelets. This technique removes the demerits of the DWT and performs superior in image denoising applications than traditional linear processing (such as wiener filtering), stationary wavelet transform (SWT), dual-tree DWT, double-density DWT etc. In this paper, the prominent results in terms of PSNR, MSE and Histogram of the proposed system are compared with dual-tree complex wavelet transform and global thresholding method. From experimental point of view, the grayscale images are considered which are corrupted by Gaussian noise.