Vision Based Interior Intrusion Detection System For Vehicles

With the rapid development of the automotive industry,vehicle population increase rapidly,and the crime about vehicle theft sustainable growth.The existing anti-theft system research focused on the stealing of whole vehicle and mainly preventing theft from entering the vehicle,but the belongings in the vehicle are often unprotected.On the other hand,the anti-theft systems on the market generally do not perform well in power consumption,standardize or alarm accuracy rate.This paper proposed a vision-based intrusion detection system for vehicles.The system includes vision subsystem and pre-warning subsystem.The vision subsystem is used to detect the window state and interior intrusion of the vehicle.To reduce the power consumption,a low-power pre-warning subsystem is designed to detect whether the windows are smashed or not.Then the system can decide whether power up the high-power-consumption visual subsystem.At the same time,the pre-warning subsystem can detect the tires stolen and the abnormal traction.The experiment results show that the system has a high detection rate,low power consumption and extraordinary suitable for the automotive application.The main work and achievements of this paper are as follows:1.Proposed a fully functional anti-theft system for vehicles.The system can not only detect the whole car stealing(abnormal traction),tire stealing but protect the belongings in the car by surveilling the window state.2.Designed a complete hardware of vision based interior intrusion detection system of vehicles.In order to meet the requirements of low power consumption,our system is divided into two embedded collaborative subsystems: vision subsystem and pre-warning subsystem.To provide a good performance in standardizing and low-power consumption,a LIN interface is reserved to communicate with ECU.When selecting the chips,we firstly meet the requirement of AEC-Q100 and choose the one with the lowest power consumption.The test results show that the hardware system has low power consumption and good stability.3.Having completed the algorithm design and functional testing.On the pre-warning system,a set of acceleration data processing algorithm is designed to accomplish the detection of tires stealing,abnormal traction and smashing the window.A frame difference based method is designed to detect interior intrusion,which has high detection accuracy in the complex environment such as different light condition,pedestrian interference,and shadow.The test results show that the proposed algorithms are suitable for the embedded platform and the detection accuracy for each function is good.