According to an e4engineering story, Intrigue Technologies, a commercial spin-off of the CMU Robotics Insititute, has developed an improved method of image processing for robot vision that greatly reduces the effects of arbitrary, real-world illumination. It allows a robot vision system to use a computational model of brightness perception that is similar to that of the human eye. At present a software simulator is available but they're hard at work on an imaging chip that incorporates the algorithm. At first glance, the results look like those one could obtain with typical image processing methods such as gamma compression or histogram equalization using the Gimp but there are apparently some subtle differences. The work was done by Vladimir Brajovic and other CMU researchers at the Computational Sensor Lab. A more detailed account of how it all works can be found in a research paper by Brajovic titled, Brightness Perception, Dynamic Range and Noise: a Unifying Model for Adaptive Image Sensors (PDF format).