As assessment of crack width with precision of the order of 10-1 mm is required in periodical crack inspection of concrete structures, crack measurement has been performed by using crack-scale manually and/or by using image processing with close range photograph. In the image processing especially, closer photograph images are required for more precise measurement of cracks, which usually have width of about 0.1 mm, because of the restriction of inherent resolving power of imaging devices.
This paper presents a crack measuring algorithm in which images of different resolution are generated from a closer range photograph and hierarchically structured from fine to coarser in order to extract concrete crack automatically by using a single spatial filter rather than using a series of spatial filters. And it is also designed to offer processing time saving from data exploding of closer range imaging with high resolution.
As a result of experiments, more than 80% of cracks above 0.1mm width are extracted automatically and RMS. error in measuring crack width is performed less than 0.19mm. It is concluded that our approach to crack measurement shows much possibility of a practical crack measuring system, which can be substituted for a conventional measurement in human-dependent inspection.
