Robust vanishing point estimation for driver assistance

This paper presents an architecture for real-time vanishing point estimation for driver assistance applications. It consists of a data-driven estimation and a model-based filtering module. The data-driven estimation algorithm is based on line-segments that are assumed to be calculated in an independent preprocessing stage. Model-based filtering is achieved by a Kalman filter that operates on the results of the data-driven processing step. The robustness of the overall estimation is significantly increased by online adaptation of the parameters of both, the data-driven as well as the model-driven processing units. The design of the feedback loop assures that no instable system states occur. The resulting architecture provides robust vanishing point estimation in a wide variety of environmental conditions