Gray Scale Image Denoising using PCA-SPT

Abstract

This paper proposed a spatially adaptive image denoising scheme, which is comprised of two stages. In the first stage, image is denoised by using Principal Component Analysis (PCA) with Local Pixel Grouping (LPG). LPG-PCA can effectively preserve the image fine structures while denoising. In the second stage, we use Steerable Pyramid Transform (SPT) to decompose images into frequency sub-bands. The noise level is updated adaptively before second stage denoising. Steerable Pyramid Transform in the second stage further improves the denoising performance. This paper also reviews on the present denoising processes and performs their comparative study. Experimental results demonstrate that the proposed PCA-SPT algorithm achieve competitive outcomes. PCA-SPT works well in image fine structure preservation, compared with state-of-the-art denoising algorithms.

Keywords: AWGN; Wavelet; SPT; LPG-PCA; BM3D; Edge preservation.