Vehicle Tracking Through Vision-Millimeter Wave Doppler Shift Fusion
Document typeMaster thesis
Rights accessOpen Access
Vehicular communications is a topic of interest in the race towards autonomous cars. Parallel, there is a future need to move towards higher frequencies in vehicular communications. In the same way, computer vision is becoming every time more used in automotive applications. In this thesis, I propose a system capable to track the position of a given car on a road with a camera and a wireless radio link in the V-band. For this I have developed some computer vision and image segmentation approaches with Gaussian mixture model for background estimation, erosion and epipolar geometry. Also, from a Vband wireless link and a signal analyser, a Doppler shift estimate has been obtained from a measurement campaign. Finally, a Kalman filter has been implemented for tracking the car?s position with a highly accurate performance.
Caracterització de la propagació a la banda de 60 GHz