基于嵌入式Linux的结构化道路车道线识别系统的研究  

Study on Line Recognition of Structured Road System Based on Embedded Linux

作　　者：邹自明[1,2] 刘晓阳[2] 

机构地区：[1]桂林理工大学广西矿冶与环境科学实验中心,桂林541004 [2]桂林理工大学机械与控制工程学院,桂林541004

出　　处：《计算机测量与控制》2014年第4期1260-1262,1266,共4页Computer Measurement &Control

基　　金：广西矿冶与环境科学实验中心资助项目(KH2011YB014)

摘　　要：结构化道路车道线检测作为车辆安全驾驶的一个重要研究方向,可在车辆偏离航道时发出报警信息,有效地抑制事故发生;现有的解决方案多是基于PC平台,但PC平台体积大,功耗高,不便运用于车载系统;为此提出了基于嵌入式Linux平台的结构化道路车道线检测方案,首先根据V4L2设计图像采集程序,图像经预处理后又分为直道和弯道两种情况进行识别,直道采用Hough变换,弯道采用贝塞尔曲线拟合,最终调用FrameBuffer驱动接口编写显示程序;方案移植到TQ6410开发平台,在城市道路上进行实验测试,平均每帧图像采集处理显示耗时在45ms以内,较为准确地检测出车道线的位置,具有较强的实时性。the line recognition of structured road is an important research direction of vehicle safety driving, when the vehicle is off course, which can send alarm information to suppress the occurrence of the accident. In the recent, The most of solutions are based on the PC platform which has some disadvantages,, for example, bulky, high power consumption and so on. The paper presents a solution which is based on embedded Linux platform. At first, according to V4L2, we design the image acquisition program. Secondly, the image acquisition is preprocessed, then we use Hough transform if the line of structured road is straight, otherwise use Bezier curve fitting. Eventually, the Framebuffer drier interface is called to display. The design is ported to TQ6410 development platform, experimental tests were carried on ur- ban roads, which the average frame is controlled in less than 45 ms and can detect the position of the drive line accurately. The plan has real --time.

关 键 词：车道线检测 Linux V4L2 FRAMEBUFFER HOUGH变换 贝塞尔曲线 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]