Design And Development Of Video Based Vehicle Abnormal Behavior Detection System

Recent developments in economic have improved people’s living standards,and resulted in the dramtic increasing of the number of motor vehicles meanwhile.Vehicles can definitely bring great conveniences to our lives,however,they also bring dangers to us.Traffic accidents occur every day all over the world,especially on the highway.It is urgent that an efficient intelligent transportation system should be built to reduce the incidence of traffic accidents.As an important part of the intelligent transportation system,vehicle abnormal behavior detection has very important significance and great practical values.This thesis is to design a vehicle abnormal behavior detection system.The works include hardware platform,software and algorithm designing and implementation.The input of the system is real time video stream,and it increases the burden of computing capability.To increase the efficiency,we design a "double DSP + double FPGA"based system.We select two ADI’s ADSP TS201 Tiger SHARC series chips as the core processors and two Xilinx’sSpartan-6 series FPGA chips XC6SLX100-3CSG484I as the core of the data communication and interface control.This thesis uses the frame difference method to detect the target vehicles,and track them by using a method based on the target region tracking.The system mainly detects three kinds of vehicle abnormal behaviors which are speeding,retrogradation and illegal parking.The results show that the proposed intelligent vehicle abnormal behavior detection system can detect vehicle abnormal behaviors effectively.