基于平行线的室内视觉导航

根据一组空间直线上的无穷远线素集在视平面上所形成的消失点对直线的平移具有稳定性的特点，提出了基于平行线的摄像机参数标定和自主移动平台室内视觉导航算法．在对摄像机进行标定时，将摄像机模型简化成关于移动平台航向角和X方向距离的线性模型，并利用走廊左右踢脚线在视平面上的投影直线的斜率、消失点坐标来标定摄像机的内、外参数；在视觉导航时，视走廊左右踢脚线为一组平行线，由其在视平面上的投影直线的斜率、消失点坐标，控制自主移动平台行驶的X方向距离和航向角，实现平台的室内视觉导航．本文采用YIQ彩色模型分割楼道图像，由Hough变换与最小二乘法相结合的方法提取楼道踢脚线．

Indoor Visual Navigation Based on Parallels

According to the stability of vanishing points of the projected lines on the image of a group of lines at infinity, the algorithms for camera calibration and indoor visual navigation of the autonomous mobile platform based on parallels are proposed. A simplified linear camera model with respect to the navigation angle and distance along Axis X of the autonomous mobile platform is established for camera calibration. Meanwhile, the gradient and coordinates of vanishing points of the projected lines on image of two baselines of the laboratory corridor are used to calibrate the inner and outer parameters of the camera model. During mobile platform visual navigation, the two baselines are viewed as the parallels, and the coordinates of vanishing points of the baselines on the image are adopted to adjust the X direction distance and course angle, as well as the gradient of baselines on the image. The color images of the corridor are segmented by model YIQ, and its baselines are detected by the combination of Hough transform and least-square method.