Abstract:

An automation extraction method for the pavement cracking with joints is propose d. It includes image preprocessing, cracking extraction algorithm based on segment connecting and joint filtering algorithm based on Hough transform. In image preprocessing, partially overlapped subblock histogramequalization (POSHE) is used to enhance the contrast of the image, it is more locally adaptive than that of the conventional histogram equalization. The binary image is obtained by a method called “localglobal matching” and high frequency emphasis. In cracking automation extraction, an automatic recognition algorithm based on connected domain correlating is proposed, it joints correlated cracking segments as one and extracts the cracking in pavement image with lots of texture noise. The result of cracking detection is finally obtained by removing joint on pavement, that is based on Hough transform. Experimental results demonstrate that the proposed cracking recognition method can accurately detect cracks on cement pavement with jint, and meets the requirements of the realtime processing..

Key words: pavement cracking recognition, image analysis, connected domain correlating, Hough transform