NCTU

Prediction of Approaching Vehicles to Blind Spots by Using a Rear-mounted Fisheye Camera

2017/08/02
CodeX16-DTL-00-1
Professorprofessor Dalton Lin
ApplicationVehicle
FunctionBehavior analysis, Image detection, Image recognition, and Object/human tracking
Technical BenefitCost-reduction
Technology StatusCan be transferred

Drivers require a safe and convenient driving environment when they are driving. Con- temporary vehicle safety systems have rapidly improved and gained public attention. The proposed system receives real-time images from a rear-mounted sheye camera and assists drivers in avoiding many dangerous conditions. Research indicates that driving accidents of- ten occur when drivers are switching lanes. Therefore, to improve driving safety, this study developed a system for predicting vehicle trajectory. When accidents are likely and risks are present, the system alerts the driver. An AdaBoost algorithm was employed for vehicle detection, and a Kalman lter was selected as the main component of the proposed system; it is an eective means for predicting the trajectories of approaching vehicles in the blind spots to prevent collisions from behind during lane switching.

1.Prediction of vehicle

2.Warning for collision situation

demo video:

Spec. request

platform:  Window 7

tools:  C/C++

hardware specification:  Intel I5

system demand: Ram: at least 4GB