Enhanced Residual Dense Intrinsic Network for Intrinsic Image Decomposition

Intrinsic image decomposition is a challenging task, which aims at recovering intrinsic components from the observation. Hand-crafted priors have been widely used in traditional methods, yet with unsatisfactory performance of quality and runtime. Recently, network-based approaches have been greatly developed, but the physical imaging principle is ignored causing the multiplication of estimated components is hard to reconstruct the observation. To overcome these limitations, we develop an enhanced residual dense intrinsic network (ERDIN) for intrinsic decomposition. Specifically, we construct the basic module (i.e., enhanced residual dense block (ERDB)) to fully exploit the hierarchical features. The physical imaging principle is designed as the reconstruction loss to ensure the consistency between the observation and the multiplication of estimated components, which is of equal importance with the data loss. Extensive experimental results illustrate our excellent performance compared with other state-of-the-art methods.

• Publication year

  2019

• Venue

  IEEE International Conference on Multimedia and Expo

• Publication date

  2019-07-01

• Fields of study

  Computer Science

• Identifiers
  DOI 10.1109/ICME.2019.00253
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Learning Intrinsic Image Decomposition from Watching the World

  2018cited by this paper
• Residual Dense Network for Image Super-Resolution

  2018cited by this paper
• Revisiting Deep Intrinsic Image Decompositions

  2017influential reference
• DARN: a Deep Adversial Residual Network for Intrinsic Image Decomposition

  2016cited by this paper
• Learning Non-Lambertian Object Intrinsics Across ShapeNet Categories

  2016cited by this paper
• Densely Connected Convolutional Networks

  2016cited by this paper
• Reflectance Adaptive Filtering Improves Intrinsic Image Estimation

  2016influential reference
• Learning Ordinal Relationships for Mid-Level Vision

  2015cited by this paper
• Learning Data-Driven Reflectance Priors for Intrinsic Image Decomposition

  2015influential reference
• An L1 image transform for edge-preserving smoothing and scene-level intrinsic decomposition

  2015influential reference
• Direct Intrinsics: Learning Albedo-Shading Decomposition by Convolutional Regression

  2015influential reference
• Learning lightness from human judgement on relative reflectance

  2015cited by this paper
• Intrinsic images in the wild

  2014influential reference
• A Simple Model for Intrinsic Image Decomposition with Depth Cues

  2013cited by this paper
• A Naturalistic Open Source Movie for Optical Flow Evaluation

  2012influential reference
• A Closed-Form Solution to Retinex with Nonlocal Texture Constraints

  2012cited by this paper
• Intrinsic Images by Clustering

  2012cited by this paper
• Intrinsic images decomposition using a local and global sparse representation of reflectance

  2011cited by this paper
• Intrinsic images using optimization

  2011cited by this paper
• Ground truth dataset and baseline evaluations for intrinsic image algorithms

  2009influential reference
• Estimating Intrinsic Images from Image Sequences with Biased Illumination

  2004cited by this paper
• Deriving intrinsic images from image sequences

  2001cited by this paper
• RECOVERING INTRINSIC SCENE CHARACTERISTICS FROM IMAGES

  1978cited by this paper
• Lightness and retinex theory.

  1971cited by this paper
• Ieee Transactions on Pattern Analysis and Machine Intelligence Shape, Illumination, and Reflectance from Shading

  year unknowncited by this paper

• Table of Contents

  2022cites this paper
• Intrinsic Image Decomposition by Pursuing Reflectance Image

  2022influential citation