基于单目视觉的车辆行驶方向检测及误差分析

智能车辆中一个重要的技术就是如何准确计算车辆前进方向与道路的夹角,目前的参考文献中基本上都是假设相机光轴与地面平行,而忽略了相机的安装误差,本文基于单目摄像机利用机器视觉方法与几何方法完成了对车辆行进方向与道路夹角的检测,给出了完整的技术实现方案,并重点分析了摄像机光轴方向与地面成一定角度情况下的校正方法,从而减少了摄像机安装误差对测量结果的影响,拓展了道路夹角检测技术的应用范围,使其有了更广的适用性,实际拍摄的场景图像也验证了本文算法的有效性。 

Moving Orientation Detection of Vehicles and Error Analysis Based on Monocular Vision

An important technology in the area of Intelligent Vehicle(IV) is how to accurately calculate the angle between the moving orientation of the vehicle and the driveway, most of the literatures nowadays have a hypothesis that the optical axis of the camera is parallel to the ground, so the installation error of the camera has been neglected. This paper achieved the angle detection between the moving orientation of the vehicle and the driveway based on monocular vision, detailed technical realization scenario has been given, and we emphatically analyzed the rectification method under the circumstance of there is an angle between the optical axis of the camera and the ground, by using our method, the effect of the installation error towards the detection results can be greatly decreased, so this paper has broaden the application scope of angle detection technique and has made it become more robust, images captured from the real scene also validated the accuracy of our method.