Research On Lane Departure Warning System Based On Image

Lane departure warning, as one of the key technologies in the field of automotiveactive safety, can effectively avoid the traffic accidents caused by the unconsciousveering off. According to statistics, accidents caused by the driver dozing orconcentration lane departure accounted for about one-third of the traffic accident.Therefore, lane departure warning system research has an important significance toensure transportation safety and reduce traffic accidents.In this paper, relying on the independent research and development program of theState Key Laboratory of Advanced Design and Manufacturing of Automotive Body ofHunan University, a real-time lane departure warning system based on image isstudied. The main work is described as follows:1. The whole system solutions and hardware design: first, comparing theadvantages and disadvantages of existing technology solutions, lane departurewarning system based on vision is designed in this paper. And this paper has used thecombination of FPGA (Field Programmable Gate Array) and DSP (Digital SignalProcessor) as the hardware architecture of the system. Then, the chip selection and thewhole system hardware design are completed in this paper. And the Cyclone of Alteracorporation series III EP3C16F484FPGA chip and TI (Texas Instruments) company’slatest video processor TMS320DM642DSP chip are used in this paper.2. A lane departure warning algorithm based on reference line: When designingsystem acquisition module, this paper adopts a CCD (Charge Coupled Device) camera,in which a reference line are embedded, as the front end of the imaging system. Basedon the reference line, a new lane departure warning algorithm is proposed in thispaper. The image preprocessing algorithm are firstly carried out on the road imagecollected by camera such as Grayscale processing, median filtering and Sobel edgedetection, and then Hough transform is used to identify the current vehicle lane line,finally by comparing the parameters of the current vehicle lane line and that of thereference line to determine the deviations from the vehicle.3. Algorithm transplantation: This paper adopts CCS3.1(Code Composer Studio3.1) as the development platform, and the transplantation of lane departure warningalgorithm has been completed based on DSP/BIOS (Digital Signal Processor/BasicInput Output System) real-time operating System. 4. System model building and experiment: In this paper, by constructing a simplesystem model, system verification experiment was carried out. And the feasibility oflane departure warning system based on image is verified by the experiments.