Design of Natural Image Denoising Filter Based on Second-Generation Wavelet Transformation and Principle Component Analysis

In this study, a novel image de-noising technique is suggested by using combination algorithms. This algorithm integrates the wavelet based de-noising algorithms and the statistical approach principle component analysis (PCA) in order to develop the subjective and objective quality of the observed image that resulted from the filtering processes. We utilize the important properties in the second-generation wavelet based de-noising and PCA to achieve better performance from our designed filter. One of the pivotal advantages of the secondgeneration wavelet transformation in the field of noise suppression is its ability to keep the main features of the signal energy in small amount of coefficients in the wavelet domain. Furthermore, one of the main features of PCA is that the energy of the signal concentrates on a very few subclasses in its domain, while the noise's energy spreads fairly among the entire signal; this characteristic allows us to separate the noise form the original signal perfectly. Our technique compares favorably against several state-of-the-art filtering systems algorithms, such as Contourlet soft thresholding, Scale mixture by wavelet transformation, BM3D, and Normal shrink. In addition, the proposed algorithm attains competitive efficacy compared with the traditional algorithms, mainly when it comes to investigating the problem of how to preserve the small details and the structures of the reconstructed image such as the edges and flat regions. On the other hand, our algorithm performed well in terms of the computational complexity where it took around 3-4 seconds to perform the de-noising operations.