Research Of Lane Departure Warning System Based On FPGA

With the development of economy and transportation, the automobile has become anindispensable part in people’s life. Though it has bring comfort and convenience to thelife of people, a series of traffic accidents has taken place with it, and a large number ofcasualties and economic losses has been caused. People felt obliged to pay more attentionto the security issues. In recent years, the active safety technology of automobile hasbecome a hotspot to researchers. As an important part of vehicle active safety technology,Lane departure warning technology has been significance to reduce the traffic accidentswhich caused by changing lane abnormally.In this paper research on the lane departure warning system, including road imageacquisition system based on FPGA, algorithm of road marking identification and thejudgment of vehicle deviates.The hardware of the system based on EP3C25F324C8N FPGA chip which productedby company of Altera. First, collect road image in CVBS format through CCD camera,the infromationg of image transform into YCrCb4:2:2format through ADV7180videoconversion chip.Then, the digital image transform into24bit RGB format in FPGA chipfor further processing. Finally, the digital image will be converted to analog formatthrough SDRAM data cache module and ADV7123, and be displayed in LCD screen.The design of software is divided into two parts: road marking identification and thejudgment of vehicle deviates. First, utilize kalman filter to remove noise. Then usedual-threshold segmentation to extract the pixel information of white and yellow in theimage. After that, establish dynamic ROI in some parts of image’s pavement section.Secondly, use PCA method to reduce the dimension of road marking. Then extractmarking feature by using Canny edge detection operator, and use the method of SVM toidentify marking information, then to carry on the fitting through the Hough transform.Finally, this paper estimate whether the vehicle deviates from the driveway according theangle of the lane markings and horizontal direction in the image.The experimental of this paper calibrate the camera at first, and then, prove theadvantage of dual-threshold algorithm and SVM method in extracting edge of themarking and identifying the characteristics of line through comparison. Last, prove theeffect of algorithm of road marking identification by using Matlab. Experiment can provethat the algorithm in this paper can achieve a good recognition effect.