Outdoor Illumination Estimation for Mobile Augmented Reality: Real-Time Analysis of Shadow and Lit Surfaces to Measure the Daylight Illumination

A realistic illumination model in Augmented Reality (AR) applications is crucial for perceiving virtual objects as real. In order to correctly blend digital content with the physical world it is necessary to measure, in real time, the illumination present in the scene surrounding the user. The paper proposes a novel solution for real-time estimation of outdoor illumination conditions, based on the video stream from the camera on handheld devices. The problem is formulated in a radiometric framework, showing how the reflected radiance from the surface maps to pixel values, and how the reflected radiance relates to surface reflectance and the illumination environment. From this we derive how to estimate the color and intensity of the sun and sky illumination, respectively, using areas in the video stream that are in direct sunlight and in shadow. The presented approach allows for rendering augmentations that adapt in real-time to dynamically changing outdoor illumination conditions.