Denoising method selection by comparison-based image quality assessment
Haoyi Liang, Daniel S. Weller · 2016
Based on a diverse range of priors on natural scene images and noise, numerous denoising algorithms have been proposed in the literature. The image quality resulting from different denoising algorithms may vary significantly across a data set. In this work, we propose a denoising algorithm selection framework that chooses among different denoising algorithms using comparison-based image quality assessment. Extensive experiments on two databases show that the proposed comparison-based selection framework consistently selects the SSIM-optimal denoising algorithm without a reference image. The proposed selection method effectively removes the burden of selecting a denoising method for applications involving processing large data sets automatically.