Locating Informative Bright Region in tunnel scenes using lighting and traffic lane cues

To locate Informative Bright Region (IBR) in which visual information is missing due to limited dynamic range of image sensor, an algorithm is proposed via combining geometric properties of visual cues into a confidence map. Location of IBR in road tunnel scenes is estimated in real-time under the condition that most of vision information inside IBR is lost. It is evaluated by comparing the estimated location of IBR with that annotated by multiple human observers on a self-built tunnel scene video dataset recorded by a car-mounted camera, and running time of the algorithm on each frame achieves 10ms. The algorithm aims at providing control timing of imaging sensor on a low cost platform like vehicle driving recorder to enhance the captured visual contents in over-exposed regions.