High-Speed, High-Resolution 3-D Machine Vision for Automated Inspection of Highway Pavement Surfaces

高速公路系统对美国的经济至关重要。根据联邦公路管理局进行的调查，90%的个人旅行和25%的货运量是由高速公路提供服务的。1998年，联邦、州和地方政府预计将在高速公路上总共花费超过1060亿美元。其中很大一部分将用于系统保护，特别是路面3R(重建、修复和重铺路面)。路面状况检测在路面3R中起着举足轻重的作用。目前，这些检测大多是通过人工目测的穹顶。这是一种缓慢、劳动强度大、繁琐、危险、昂贵和主观的方法。为了解决这些问题，需要开发一套自动化检测系统。在这项研究中，我们提出了一种新的用于路面自动检测的三维机器视觉技术。该技术基于彩色编码条纹投影和相移概念，可提供高速(仅受相机的帧速率限制)和高分辨率(仅受相机中的像素数限制)的3-D曲面轮廓绘制能力。该系统与专门开发的车辆集成在一起，可以在正常高速公路速度(每小时55英里)下提供路面状况的实时检测。利用基于小波分析的先进算法，可以实现对各种路面病害的自动检测、识别和测量。此外，如果使用全球定位系统(GPS)和地图软件，就有可能精确定位路面破损的位置，并在必要时带领工人的维修工程师前往工地。主要研究内容如下：1.提出了一种基于彩色编码条纹投影和相移的三维机器视觉技术，用于高速、高分辨率路面检测。2.开发了一种结合了三维机器视觉系统和GP5的专用车辆，用于高速公路速度下的路面数据自动采集。3.提出了基于小波分析技术的路面病害自动检测与定量评价算法。

The highway system is vital to the economy of the United States. According to the survey conducted by the Federal Highway Administration, 90 percent of personal travel and 25 percent of freight movement is served by highways. In 1998, federal, state, and local governments are expected to spend a combined sum of over $106 billion on highways. A significant portion of it will be used for system preservation, in particular, pavement 3R (reconstruction, rehabilitation, and resurface). Inspection of pavement surface conditions plays a pivotal role in pavement 3R. Currently, these inspections are dome mostly by manual visual inspection. This is a slow, labor-intensive, tedious, dangerous, expensive, and subjective method. To solve these problems, an automated inspection system needs to be developed. In this research, we propose to develop a novel 3-D machine vision technique for automated inspection of pavement surfaces. This technique is based on a color-encoded fringe projection and phase shifting concept that provides high-speed (limited only by the frame rate of the camera) and high-resolution (limited only by the number of pixels in the camera) 3-D surface contouring capability. Integrated with a specially developed vehicle, the system can provide real-time inspections of pavement surface conditions at normal highway speeds (55 mph). With advanced algorithms developed based on wavelet analysis, automatic detection, identification, and measurment of various types of pavement distresses can be realized. Moreover, if global positioning system (GPS) and a map software are used, it is possible to pinpoint the locations of pavement distresses and if necessary, lead the repair engineerrs of workers to the sites. The objectives of the proposed research are as follows: 1. Develop a novel 3-D machine vision technique based on color-encoded fringe projection and phase shifting for high-speed, high-resolution pavement inspection. 2. Develop a special vehicle that incorporates the 3-D machine vision system and GP5 for automated pavement data collection at highway speeds. 3. Develop algorithms for sutomatic detection and quantitative evalutation of pavement surface distresses based on wavelet analysis techniques.