IDEA: Intrinsic Decomposition Algorithm With Sparse and Non-local Priors

World’s FIRST 10K Best Paper Award – Platinum Award

ABSTRACT

This paper proposes an new intrinsic decomposition algorithm (IDEA) that decomposing a single RGB-D image into reflectance and shading components. We observe and verify that, shading image mainly contains smooth regions separated by curves, and its gradient distribution is sparse. We therefore use L1-norm to model the direct irradiance component--the main sub-component extracted from shading component. Moreover, non-local prior weighted by a bilateral kernel on a larger neighborhood is designed to fully exploit structural correlation in the reflectance component to improve the decomposition performance. The model is solved by the alternating direction method under the augmented Lagrangian multiplier (ADM-ALM) framework. Experimental results on both synthetic and real datasets demonstrate that the proposed method yields better results and enjoys lower complexity compared with the state-of-the-art methods.

Keywords: Intrinsic decomposition, RGB-D, sparse, non-local

SOURCE CODE 

Opening soon. The source code is only for the non-commercial use.

PUBLICATIONS

[1] Yujie Wang, Kun Li, Jingyu Yang, Xinchen Ye, “Intrinsic image decomposition from a single RGB-D image with sparse and non-local priors”, IEEE International Conference on Multimedia and Expo (ICME),.July 10-15, 2017, Hongkong, China. [pdf]

[2] Kun Li, Yujie Wang, Jingyu Yang, Xinchen Ye, “IDEA: Intrinsic decomposition algorithm with sparse and non-local priors”, submitted to IEEE Transactions on Image Processing, 2017.