Image debanding using cross-scale invertible networks with banded deformable convolutions.

Journal: Neural networks : the official journal of the International Neural Network Society
PMID:

Abstract

Banding artifacts in images stem from limitations in color bit depth, image compression, or over-editing, significantly degrades image quality, especially in regions with smooth gradients. Image debanding is about eliminating these artifacts while preserving the authenticity of image details. This paper introduces a novel approach to image debanding using a cross-scale invertible neural network (INN). The proposed INN is information-lossless and enhanced by a more effective cross-scale scheme. Additionally, we present a technique called banded deformable convolution, which fully leverages the anisotropic properties of banding artifacts. This technique is more compact, efficient, and exhibits better generalization compared to existing deformable convolution methods. Our proposed INN exhibits superior performance in both quantitative metrics and visual quality, as evidenced by the results of the experiments.

Authors

  • Yuhui Quan
    School of Computer Science and Engineering, South China University of Technology, Guangzhou, 510006, China. Electronic address: csyhquan@scut.edu.cn.
  • Xuyi He
    School of Computer Science and Engineering, South China University of Technology, Guangzhou, 510006, China. Electronic address: csxuyihe@mail.scut.edu.cn.
  • Ruotao Xu
    Institute for Super Robotics, South China University of Technology, Guangzhou, China; Key Laboratory of Large-Model Embodied-Intelligent Humanoid Robot, Guangzhou, China. Electronic address: rtxu@superobots.com.
  • Yong Xu
    Department of Psychiatry, The First Hospital of Shanxi Medical University, Taiyuan, China.
  • Hui Ji
    Department of Mathematics, National University of Singapore, Singapore.