Authentication based on face recognition can fail under poor lighting or exposure conditions. This disclosure describes techniques to improve the brightness and/or contrast of a given facial image such that face recognition based authentication can be performed robustly over a wide range of lighting or exposure conditions. Brightness is improved using gamma correction, where the optimal gamma value is given by a low-complexity, closed-form formula. Contrast is improved by subjecting pixel intensities to a mapping that results in better color saturation, or a larger standard deviation of pixel intensities. The techniques work on near infra-red (NIR) as well as visible-spectrum (RGB) images.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
N/A, "Face Image Normalization For Authentication Under Diverse Lighting Conditions", Technical Disclosure Commons, (July 09, 2020)