An Optical physics inspired CNN approach for intrinsic image decomposition

TL;DR

This paper introduces a physics-inspired CNN model for intrinsic image decomposition that effectively separates reflectance and shading from a single image without ground truth, outperforming existing methods.

Contribution

It presents a novel neural network architecture leveraging physics-based parameters for unsupervised intrinsic image decomposition, addressing a key gap in current deep learning approaches.

Findings

Outperforms existing deep learning IID techniques

Physics-based parameters significantly enhance decomposition quality

Provides detailed analysis and potential avenues for future improvement

Abstract

Intrinsic Image Decomposition is an open problem of generating the constituents of an image. Generating reflectance and shading from a single image is a challenging task specifically when there is no ground truth. There is a lack of unsupervised learning approaches for decomposing an image into reflectance and shading using a single image. We propose a neural network architecture capable of this decomposition using physics-based parameters derived from the image. Through experimental results, we show that (a) the proposed methodology outperforms the existing deep learning-based IID techniques and (b) the derived parameters improve the efficacy significantly. We conclude with a closer analysis of the results (numerical and example images) showing several avenues for improvement.