9DA tunnel scanner system with a function to detect damage using image processing
  • Development of an image-processing technique to automatically combine multiple images to generate a cross-sectional image
  • Automatic detection of fine cracks in typical walls
  • Development and commercialization of a low-cost image pickup device for use as a tunnel scanner
  •    The conventional method of combining multiple images to generate a developmental cross-sectional image had the disadvantage of being labor-intensive and time-consuming. To combat this, the RTRI applied the technique of mosaicking to automatically detect characteristic points based on differences in color and pattern and superimpose images based on the relationship between different points. This enabled developmental images to be generated in about half the time taken by conventional methods (Fig. 1).
       To create a commercial tunnel damage-detection system that automatically generates developmental drawings of damage, the RTRI developed a high-precision algorithm to detect cracks in tunnel walls. By applying the algorithm to actual image pickup, the RTRI found that it was possible to precisely measure the position, width, length and direction of cracks in typical walls that are free from extreme dirt (Fig. 2). This can be applied to the tunnel inspection system(Tonnelus) currently in service.
       After developing the low-cost image pickup device, the RTRI commercialized a tunnel scanner system featuring the above functions (Fig. 3).


    HOME > Major Results of Research and Development in Fiscal 2005 > I Safety/Reliability


    Copyright(c) 2006 Railway Technical Research Institute, Tokyo Japan, All rights reserved.