This study describes an effective method for detecting, tracking and classifying vehicles during night-time in order to support automotive adaptive illumination applications. The hereby described software framework, which computes the relative position, velocity and estimated class of all detected vehicles, integrates multiple processing stages. Firstly, an image segmentation using a threshold method to detect all light sources in the image. Secondly, possible pairs of head- and taillight are clustered using geometrical information. Thirdly, all detected ob-jects are tracked using a Kalman-Filter to increase resolution and robustness of the algorithm. Lastly, a method for computing distance and velocity for all classi-fied objects (e.g. cars, trucks, bikes) is presented. The system is tested to run in realtime and some results and conclusions are offered at the end
Vehicle detection, classification and position estimation based on monocular video data during night-time
2009
13 Seiten, 12 Bilder, 10 Quellen
Aufsatz (Konferenz)
Englisch
REAL TIME MONOCULAR VEHICLE VELOCITY ESTIMATION USING SYNTHETIC DATA
British Library Conference Proceedings | 2021
|Monocular Video Measurement Of Position And Orientation
NTRS | 1996
|