Bridge Crack Measurement System Research Based On Image Processing Techniques

Crack is a common disease of the concrete bridge and it is a very important work for rail maintenance department to detect, identify and repair the bridge crack. Traditional practices of bridge crack detection depends mainly on the manual work, the disadvantage of this method is high-risk, low precision, low efficiency. At present, Crack measurement system based on digital image processing technology are mostly semi-automatic or artificial-dependent with quite low efficiency. How to make this system more intelligent, more efficient, more accurate and make the effect of image processing better has important practical significance. The main innovation of this paper as follows.Firstly, due to the effects of noise, Crack image processing has low recognition accuracy, this paper come up with a suitable image segmentation method and use the morphological processing method to eliminate noiseisolated areas after image binarization. The experimental results show that the effect of crack image processing is better and the recognition accuracy is higher.Secondly, after image binarization, image crack area usually break and has gaps, this paper present a new algorithm of filling crack’s gap. Experimental investigation shows that this algorithm is a quite good solution to the problem that crack area has gaps.Finally, for the problem that the crack measurement accuracy is not high, after a comprehensive comparison of several method this paper put forward a method to obtain the maximum width which based on crack trend and edge pixels. According to its skeleton the crack arealength and average width was obtained. In this way the problem that low accuracy of geometry information calculation is solved.