Crack Recognition According to Image Acquisition Condition

초록

The crack inspection is important factor for evaluating the safety of concrete structure. Research has developed to replace present visual inspection with image-based automatic crack recognition system. The previous research mostly related to develop of algorithm for extracting crack features accurately. However, the effect of illuminace and measurement distance have to be considered to use the crack recognition system in the field. In this paper, To find out the effects, we carry out two experiments after attaching simulated cracks which describe various brightness and width on the concrete tunnel. First, to find out the effect of distance between crack and camera, we have taken photographs at difference distances. Second, to find out the effect of illumination, we have taken photographs when illumination is 9 lx, 90 lx, 960 lx, 10,700 lx. To analysis the effect of measurement distance on the crack recognition, the number of crack pixel was counted according to the measurement distance after converting original image to binary image. When the distance was increased, the number of crack pixel was decreased and it makes the crack recognition difficult. We classified ability of crack recognition into two classes according to the number of crack pixel. One is recognizable minimum crack width and the other is extractable minimum crack width. Recognizable is only able to recognize the crack and extractable is able to estimate the crack width by calculating the length per pixel. Also, we investigated the change of pixel value according different illuminance and brightness of the cracks. When the illuminance was decrease, the pixel value of crack and the background were decrease and the brightness difference between the cracks and background were decreased. As a result we estimated the minimum recognizable and extractable crack width according to the measurement distance and illuminane. The result can be used as a reference when apply image detect system to the field.