Improved Photometric Stereo through Efficient and Differentiable Shadow Estimation
Journal
34th British Machine Vision Conference, BMVC 2023
Start Page
216972
Date Issued
2023-11-20
Author(s)
Abstract
Traditional photometric stereo approaches, although valuable in various applications, have faced limitations due to lack of considering accurate shadow estimation under different object geometry and varying lighting conditions. To address this issue, we propose a fast and accurate shadow estimation algorithm based on a dynamic programming-based sampling method with a differentiable temperature function. The proposed method can be easily used to improve existing photometric stereo methods for better estimation of shadow estimation results. In addition, we further improve the performance with our proposed higher-order derivation loss configuration. To assess the effectiveness of our method, we conduct comprehensive experiments and compare our results with diverse unsupervised and supervised approaches. The results demonstrate that our method consistently outperforms other state-of-the-art unsupervised methods in terms of mean angular error (MAE) while remaining competitive with supervised techniques.
Event(s)
34th British Machine Vision Conference, BMVC 2023
Publisher
British Machine Vision Association, BMVA
Type
conference paper