Photometric Stereo by Hemispherical Metric Embedding

TL;DR

This paper introduces a novel hemisphere embedding method for photometric stereo that accurately reconstructs 3D shapes from images with diverse reflectance properties without relying on specific reflectance models.

Contribution

It presents a new embedding technique that generalizes shape reconstruction in photometric stereo to various reflectance types, outperforming existing manifold learning approaches.

Findings

Outperforms existing manifold learning methods in hemisphere embedding.

Successfully reconstructs shapes with diverse reflectances including diffuse and specular surfaces.

Achieves more accurate shape reconstructions even with shadows and complex reflectance.

Abstract

Photometric Stereo methods seek to reconstruct the 3d shape of an object from motionless images obtained with varying illumination. Most existing methods solve a restricted problem where the physical reflectance model, such as Lambertian reflectance, is known in advance. In contrast, we do not restrict ourselves to a specific reflectance model. Instead, we offer a method that works on a wide variety of reflectances. Our approach uses a simple yet uncommonly used property of the problem - the sought after normals are points on a unit hemisphere. We present a novel embedding method that maps pixels to normals on the unit hemisphere. Our experiments demonstrate that this approach outperforms existing manifold learning methods for the task of hemisphere embedding. We further show successful reconstructions of objects from a wide variety of reflectances including smooth, rough, diffuse and specular surfaces, even in the presence of significant attached shadows. Finally, we empirically prove that under these challenging settings we obtain more accurate shape reconstructions than existing methods.