Real-Time Vehicle Classification using a Smart Embedded Device with a `Silicon Retina' Optical Sensor

This paper presents a sensor system for traffic data acquisition capable of discriminating vehicles into two classes (cars and trucks). The system is based on a bio-inspired `silicon retina' optical sensor in a compact embedded system and has been exhaustively evaluated in a real world environment at a highway test site. We further present the classification approach, data from the evaluation of the system under different lighting conditions and analyze sources of classification errors. The System achieves a 94.4% classification rate for cars and 92.2% classification rate for trucks under average daylight conditions.