Algorithms for a real-time HDR video system

When the dynamic range of radiance values in a scene exceeds the capabilities of a camera, a single picture can only capture one brightness range of the scene faithfully at a time. We propose a system for creating high dynamic range (HDR) videos that overcomes this limitation. It acquires a number of images under varying exposure settings from dark to bright, each containing new scene radiance information. The camera motion between the images is compensated and they are fused into a single HDR frame. For visualization on regular displays, the video frame is tone mapped to the output range of the display. We introduce algorithms for reduced redundancy acquisition, efficient registration and visualization of HDR video that are fast enough to be used in real-time.