Bio-Inspired Motion Detection for a Blind Spot Overtaking Monitor

S. Mota, E. Ros, E.M. Ortigosa, and F.J. Pelayo


Reichardt correlator model, motion detection, segmentation, blind spot


The rear-view mirror is unhelpful when an overtaking car is in the blind quadrants (blind spot). In this contribution we describe the software implementation of an algorithm to monitor vehicle- overtaking processes. This algorithm detects the vehicle to the rear and determines whether it is approaching or not; if approaching, the algorithm alerts us of the vehicle’s presence. The proposed system is based on the Reichardt correlator model. The approach presented uses the saliency of motion features in a competition scheme to filter noise patterns. In this way features corresponding to rigid body motion self-emerge from the background. Real overtaking sequences have been used to develop this monitoring system.

Important Links:

Go Back