Research on crack detection algorithm of asphalt pavement

A novel, efficient image processing method is proposed here for extraction of pavement cracks from fuzzy and discontinuous pavement images. Pavement surface images obtained by CCD array, where pavement cracks are often blurry and discontinuous due to particle materials of pavement surface, crack degradation and unreliable crack shadows. Firstly, A series of preprocessing including using histogram specification, dealing with Canny-HBT filter, brightness non-uniformity correction and contrast enhancement are implemented, as the aim is to enhance the differences between cracks and backgrounds; Then a multi-scale curvelet transform is presented to obtain multiple resolution image representation, using multi-resolution method which could reserve more geometric characteristics of images, and the reliability and precision of the crack detection were improved to a large extent; Finally, the use of the Max-Mean fusion operations which consist of the dilation, erosion and thin are performed on the obtained image, and making threshold decision on the cracks connected area to realize the fine cracks fusion. The proposed system which can effectively extract the small cracks and the cracks with the weak contrast, and it has strong robustness and practical value, gives improved edge detection in images with superior edge localization and gains higher PSNR.