INSPECTION METHOD BASED ON IMAGE PROCESSING FOR MEASURING LEVEL DIFFERENCE OF MANHOLE COVERS
Daisuke Uchibori1, Hiroyuki Takahashi2, Shoji Mochizuki3 and Kazuyoshi Kawabata4
1) Engineer, Civil Engineering Project, NTT Access Network Service Systems Laboratories, Nippon Telegraph and Telephone Corporation.
2) Senior Research Engineer, Civil Engineering Project, NTT Access Network Service Systems Laboratories, Nippon Telegraph and Telephone Corporation.
3) Senior Research Engineer, Civil Engineering Project, NTT Access Network Service Systems Laboratories, Nippon Telegraph and Telephone Corporation.
4) Senior Research Engineer, Supervisor, Civil Engineering Project, NTT Access Network Service Systems Laboratories, Nippon Telegraph and Telephone Corporation.
Abstract: At present, huge number of manhole covers exists on road surface. Owners of manholes inspect periodically to see if there is a level difference between a manhole cover and the frames they fit into. If there is an excess level difference between a manhole cover and the frame, it could result in an uncomfortable and noisy ride for people driving over it. The current method of measuring level difference involves the inspector setting up a working zone on the road then using a caliper to take measurements. This method is problematic in that it is necessary to apply for a permit before conducting the on-site inspection, set up a working zone to ensure the safety of the inspector, and affect local traffic for the duration of the inspection. We propose a safe and efficient method based on image processing for automatically measuring the level difference of manhole covers and an algorithm for image processing. An inspector photographs the manhole cover from the sidewalk or another safe location with a digital camera as an alternative to using a caliper. Then, the image processing algorithm finds the edge of the level difference of a manhole cover in the photograph. Edge detection is possible by taking into account the shape of the level difference and limiting unnecessary edges in an image. We verified that the edge detection with the algorithm produces a clear edge of the level difference and a smaller number of unwanted edges compared with other conventional edge detection method. We evaluated the accuracy of the proposed method in estimating the level difference by using 123 images of manhole covers on road. The proposed inspection method exhibited a maximum measuring error of 1.8 mm, which is sufficiently small for detecting dangerous level differences.
Keywords: inspection of infrastructure, manhole cover, image processing, measuring level difference
Bibliographical Reference:
Daisuke Uchibori, Hiroyuki Takahashi, Shoji Mochizuki, and Kazuyoshi Kawabata. “INSPECTION METHOD BASED ON IMAGE PROCESSING FOR MEASURING LEVEL DIFFERENCE OF MANHOLE COVERS.” In Proceedings of International Conference on Civil and Building Engineering Informatics (ICCBEI 2015), 96. Tokyo, Japan, 2015.