A color vision system for real-time analysis of road scenes

This paper briefly describes a new real-time color vision system for analyzing the road situation based on new methods for producing concise image descriptions and for segmenting images. The technique developed is weakly sensible to changes in lighting conditions and can be used without special adjustment to the conditions of a particular road scene. The results of operation of an implementation of the system are discussed and examples of processing are given. Possible practical applications to driver assistance, traffic control, and the navigation of autonomous vehicles are considered.