摘要
针对目前公路边坡位移识别方法的不足,分析了基于双目视觉的公路边坡表面位移识别方法。在边坡表面设置一定数量的监测目标点,利用2台固定的摄像机采集边坡图像,采用亚像素搜索算法提取监测点的图像坐标,通过像平面坐标系、摄像机坐标系、测量坐标系之间的转换关系,计算监测点的三维坐标,识别边坡表面位移,并进行了可行性模拟验证。结果表明:监测点的计算位移值与实际值基本吻合,相对误差为2.02%,9组样本试验的标准差为0.832mm,测点误差值小于±1,±1.5,±2mm的比例分别为65%,89%,98%。
To overcome the shortcomings of highway slope displacement identification, a new computation method based on binocular vision was proposed to identify highway slope displacement. In this method, some targets were set on slope surface, and two cameras were installed on their right positions to acquire the slope images. The image coordinates of targets were extracted using sub-pixel search algorithm, the 3D coordinates of targets and the displacement of highway slope were computed by using the geometrical relationship among the camera, the target, and the photographed image, and its validity was examined with a case study in laboratory. Results show that the computed displacement values are consistent with the actual resuhs with 2.02% relative error. The standard deviation of nine cases is 0. 832 mm, and the targets with error values less than ± 1 mm account for 65%, less than ±1. 5 mm account for 89%, and less than ±2 mm account for 98%.
出处
《中国公路学报》
EI
CAS
CSCD
北大核心
2015年第4期27-34,共8页
China Journal of Highway and Transport
基金
国家自然科学基金项目(51178193)
交通运输部建设科技项目(2009 353-344-570)
广东省交通运输厅科技项目(2010-02-051)
关键词
道路工程
边坡位移
双目视觉
位移计算
road engineering
slope displacement
binocular vision
displacement computation
