A Study of Image Recognition Technology Applying to Pavement Distress

• Main Authors: Lin-jyun Chen, 陳林君
• Other Authors: Jyh-dong Lin
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/57b4z7

碩士 === 國立中央大學 === 土木工程學系 === 102 === Road inspection is the groundwork for road maintenance. Pavement distress tends to put the safety of the road-users are at risk and impact the pavement smoothness indirectly. In addition to the efforts made by domestic road authorities in actively implementing the plan of road smoothness, R&;D of road testing equipment, which are built in vehicles, has been made at home and abroad for the past few years in order to facilitate road inspection. Image capture devices featuring the latest technologies, such as smart phones, tablets and CCD Cameras, are installed in the testing equipment to collect photographic images of the pavement. Therefore, in this study, images are used as the basis. An automatic inspection and recognition system is developed to capture pavement images. An image processing technology is then used to capture pavement distress feature, measure the distress images and judge the severity of damage with the pavement condition index. This study involves using CCD Cameras to take pictures of the pavement and recognizing the pavement distress feature out of them. The photographic images captured by the cameras in the vehicle are slant and then corrected orthographically. The straight line and lane detection is employed to detect the lane markings which are taken as the limits of the recognition scope of pavement distress feature. The image background outside of the scope is cropped. Then the pavement distress feature is captured through illumination adjustment made with image processing technology, image binary processing, and methodology used in morphology such as erosion and dilation, border removal and hole filling. Finally, the pavement distress feature is calculated with image measurements. This research result is then verified against the calibrated images. With distance and area measurements, it is determined that the distance for capturing complete pavement image for detection is 2.5 meters. The feature of potholes, manhole, longitudinal / transverse cracking is cropped out of the pavement images with image process technology. The area, length and width of the pavement distress is measured with image measuring technology. At last, the measurements are used to severity of the pavement distress, compared against the pavement condition index. This technology is incorporated into the road inspection procedure by importing the collected data for pavement distress into the Pavement Management System as the reference in assessing the pavement engineering life cycle and determining the maintenance cycles in the future.