Deep Gaussian Conditional Random Field Network: A Model-Based Deep Network for Discriminative Denoising

Citations Over TimeTop 10% of 2016 papers

Abstract

We propose a novel end-to-end trainable deep network architecture for image denoising based on a Gaussian Conditional Random Field (GCRF) model. In contrast to the existing discriminative denoising methods that train a separate model for each individual noise level, the proposed deep network explicitly models the input noise variance and hence is capable of handling a range of noise levels. Our deep network, which we refer to as deep GCRF network, consists of two sub-networks: (i) a parameter generation network that generates the pairwise potential parameters based on the noisy input image, and (ii) an inference network whose layers perform the computations involved in an iterative GCRF inference procedure. We train two deep GCRF networks (each network operates over a range of noise levels: one for low input noise levels and one for high input noise levels) discriminatively by maximizing the peak signal-to-noise ratio measure. Experiments on Berkeley segmentation and PASCALVOC datasets show that the proposed approach produces results on par with the state of-the-art without training a separate network for each individual noise level.

Related Papers

• → A method for estimation and filtering of Gaussian noise in images(2003)119 cited
• → Impulsive noise characterization and its effect on digital audio quality(2015)6 cited
• → Higher Order Spectra Denoising Method of Phase Matching Noise Estimation(2010)3 cited
• → Estimation of Measurement Accuracy of Information Signal Parameters at Simultaneous Influence of Multiplicative and Additive Non-Gaussian Noise(2019)1 cited
• → Time Delay Estimation for Noise-Like Signals Embedded in Non-Gaussian Noise Using Robust Similarity Measures(2021)