Don't Believe Anything You See: Modern Digital Photography and the Camera Imaging Pipeline
Michael S. Brown
National University of Singapore
Abstract:Modern digital photography is not about capturing faithful recordings of the light rays that enter a camera -- it is about producing perceptually optimal images. This begs the question: can camera image values be transformed to physically meaningful values, and if so, when and how this can be done? From our analysis from over 10,000 images captured from 33 different camera makes and models we have developed a new in-camera imaging model that can accurately describe how a camera maps light measures (i.e. RAW sensor responses) to the final standard RGB color output (sRGB) under various settings, including white-balance and picture styles (e.g. landscape, portrait, etc). Additionally, we show how this new imaging model can be used to build an image correction application that converts an sRGB input image captured with the wrong camera settings to an sRGB output image that would have been recorded under the correct settings of a specific camera.
Michael S. Brown is an associate professor and assistant dean in the
School of Computing at the National University of Singapore. Dr. Brown
has several several times as an area chair for all the major computer
vision conferences (ICCV, CVPR, ECCV and ACCV). He is currently an
associate editor for the IEEE Transactions on Pattern Analysis and
Machine Intelligence (TPAMI). His research interests include computer
vision, image processing and computer graphics.