Universal Photometric Stereo Network using Global Lighting Contexts
This addresses the problem of limited usability in photometric stereo for researchers and practitioners by eliminating prior lighting assumptions, though it appears incremental as it builds on existing uncalibrated methods.
The paper tackles the universal photometric stereo task, which aims to estimate surface normals for objects with diverse shapes and materials under arbitrary lighting without assuming specific physical models, by introducing a data-driven method that uses global lighting contexts and achieves state-of-the-art performance on test data.
This paper tackles a new photometric stereo task, named universal photometric stereo. Unlike existing tasks that assumed specific physical lighting models; hence, drastically limited their usability, a solution algorithm of this task is supposed to work for objects with diverse shapes and materials under arbitrary lighting variations without assuming any specific models. To solve this extremely challenging task, we present a purely data-driven method, which eliminates the prior assumption of lighting by replacing the recovery of physical lighting parameters with the extraction of the generic lighting representation, named global lighting contexts. We use them like lighting parameters in a calibrated photometric stereo network to recover surface normal vectors pixelwisely. To adapt our network to a wide variety of shapes, materials and lightings, it is trained on a new synthetic dataset which simulates the appearance of objects in the wild. Our method is compared with other state-of-the-art uncalibrated photometric stereo methods on our test data to demonstrate the significance of our method.