基于双目视觉的水泥混凝土路面错台检测方法

Cement Concrete Pavement Faulting Detection and Recognization Based on Binocular Vision

摘要:

针对水泥混凝土路面错台测量手段的不足,提出了一种基于双目视觉的错台测量方法.该方法根据双目摄像机的成像特点和水泥混凝土路面的图像特征,对接缝位置进行定位和错台计算.定位分为2个主要步骤,一是基于灰度投影的粗定位,二是在粗定位的基础上提取接缝附近图像,利用灰度投影和边缘投影精定位.由双目视觉测量系统计算相邻2块水泥板在错台附近点的三维坐标,利用这些坐标计算错台量,其中关键环节是在双目图像匹配时提出了一种基于匹配点位置估计的匹配算法.试验表明,与直尺测量结果吻合.

Abstract:

A new measurement method for faulting detection is proposed to overcome the shortcomings of traditional measurement methods for faulting detection on the cement concrete pavement.This method is realized by slab joints positioning based on the digital image processing and faulting calculating based on binocular vision,according to double camera’s imaging characters and the cement concrete pavement’s imaging feature.Slab joints positioning can be divided into approximate location based on the gray projection feature of whole image and accurate location based on the character of gray projection and edge projection around the slab joints image.The 3D coordinates of the points around the slab faulting which is used to calculate the faulting are obtained by the binocular vision measurement system.For the key step of this system,image matching adopts a new algorithm based on the estimated location of matching points.The experimental results show that measurement accuracy meets to the test results by ruler.