Incorporating Lambertian Priors into Surface Normals Measurement

TL;DR

This paper introduces a novel photometric stereo network that incorporates Lambertian priors to improve surface normal measurement accuracy on non-Lambertian surfaces, addressing errors caused by specularities and shadows.

Contribution

The method uses Lambertian priors to reparameterize neural network weights, reducing hypothesis space and enhancing surface detail reconstruction, which is a novel approach in photometric stereo.

Findings

Improved accuracy in surface normal estimation on benchmark datasets.

Effective reduction of errors caused by non-Lambertian reflectance.

Enhanced surface detail reconstruction through differential features learning.

Abstract

The goal of photometric stereo is to measure the precise surface normal of a 3D object from observations with various shading cues. However, non-Lambertian surfaces influence the measurement accuracy due to irregular shading cues. Despite deep neural networks have been employed to simulate the performance of non-Lambertian surfaces, the error in specularities, shadows, and crinkle regions is hard to be reduced. In order to address this challenge, we here propose a photometric stereo network that incorporates Lambertian priors to better measure the surface normal. In this paper, we use the initial normal under the Lambertian assumption as the prior information to refine the normal measurement, instead of solely applying the observed shading cues to deriving the surface normal. Our method utilizes the Lambertian information to reparameterize the network weights and the powerful fitting ability of deep neural networks to correct these errors caused by general reflectance properties. Our explorations include: the Lambertian priors (1) reduce the learning hypothesis space, making our method learn the mapping in the same surface normal space and improving the accuracy of learning, and (2) provides the differential features learning, improving the surfaces reconstruction of details. Extensive experiments verify the effectiveness of the proposed Lambertian prior photometric stereo network in accurate surface normal measurement, on the challenging benchmark dataset.