Robust white balance estimation using joint attention and angular loss optimization

Zhihao Li; Zhan Ma

doi:10.1117/12.2586930

Translator Disclaimer

4 January 2021 Robust white balance estimation using joint attention and angular loss optimization

Zhihao Li, Zhan Ma

Author Affiliations +

Proceedings Volume 11605, Thirteenth International Conference on Machine Vision; 116051E (2021) https://doi.org/10.1117/12.2586930
Event: Thirteenth International Conference on Machine Vision, 2020, Rome, Italy

Abstract

White balance estimation (WBE) is one of the most fundamental and crucial steps in modern Image Signal Processor (ISP). Recent years have witnessed the advancements of deep-learning based WBE. However, existing models were mostly trained on individual datasets with limited samples captured using various camera sensors, making it hard for model generalization. In this paper, we propose a novel Channel-Attention optimized U-net model, in which an angular loss is embedded, to accurately estimate the white balance. We demonstrate our approach on recently released largescale dataset “Cube Plus” captured using the same camera sensor, offering state-of-the-art performance.

Citation Download Citation

Zhihao Li and Zhan Ma "Robust white balance estimation using joint attention and angular loss optimization", Proc. SPIE 11605, Thirteenth International Conference on Machine Vision, 116051E (4 January 2021); https://doi.org/10.1117/12.2586930

ACCESS THE FULL ARTICLE