Variational probabilistic generative framework for single image super-resolution

作     者：Wang, Zhengjue Chen, Bo Zhang, Hao Liu, Hongwei 

作者机构：Xidian Univ Natl Lab Radar Signal Proc Xian 710071 Shaanxi Peoples R China Xidian Univ Collaborat Innovat Ctr Informat Sensing & Underst Xian 710071 Shaanxi Peoples R China 

出 版 物：《SIGNAL PROCESSING》 (信号处理)

年 卷 期：2019年第156卷

页      面：92-105页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 

基　　金：Thousand Young Talent Program of China, NSFC [61771361, 61671354] 111 Project [B18039] National Science Fund for Distinguished Young Scholars of China 

主　　题：Probabilistic generative model Image super-resolution Conditional prior Recognition model 

摘      要：In this paper, a general variational probabilistic generative framework parameterized by deep networks is proposed for single image super-resolution, which assembles the advantages of coding-based methods and regression-based methods. We use probabilistic generative networks to model the joint full likelihood of a pair of low-resolution (LR) and high-resolution (HR) patches which are generated from a shared latent representation. An inference model is applied to infer the stochastic distribution of the latent representation. By jointly optimizing the generative and inference models, a regression process to the distribution of the HR patch is implied during the learning phase, which provides an efficient forward mapping to accomplish the super-resolution task. We use our framework as a guidance and develop a new model called PGM-CP, with the help of an informative conditional prior and a consistent recognition model. We likewise show how three existing popular example-based SR methods can be reinvented under our framework. The effectiveness and efficiency of the proposed method is examined based on three public datasets. Experimental results demonstrate that our model is competitive with state-of-the-art approaches, especially when the image is corrupted by noise. (C) 2018 Elsevier B.V. All rights reserved.