Uncalibrated Neural Inverse Rendering for Photometric Stereo of General Surfaces

TL;DR

This paper introduces an uncalibrated neural inverse rendering framework for photometric stereo that estimates light directions and surface normals without requiring ground-truth data, effectively handling complex surfaces with interreflections.

Contribution

It proposes a novel uncalibrated deep learning approach that estimates lighting and surface properties jointly, bypassing the need for precise light directions or normals during training.

Findings

Achieves comparable or superior results to supervised methods.

Effectively models concave and convex surface parts.

Handles interreflections in complex surfaces.

Abstract

This paper presents an uncalibrated deep neural network framework for the photometric stereo problem. For training models to solve the problem, existing neural network-based methods either require exact light directions or ground-truth surface normals of the object or both. However, in practice, it is challenging to procure both of this information precisely, which restricts the broader adoption of photometric stereo algorithms for vision application. To bypass this difficulty, we propose an uncalibrated neural inverse rendering approach to this problem. Our method first estimates the light directions from the input images and then optimizes an image reconstruction loss to calculate the surface normals, bidirectional reflectance distribution function value, and depth. Additionally, our formulation explicitly models the concave and convex parts of a complex surface to consider the effects of interreflections in the image formation process. Extensive evaluation of the proposed method on the challenging subjects generally shows comparable or better results than the supervised and classical approaches.