| The detection of asphalt pavement crack is an important part of the highwaymaintenance and management. In recent years, based on image processing, technologyto do automatic crack detection has been applied widely, and the detection algorithm isdirectly affects the test results. Because of the complexity of the pavement crack image,the detection effect is not accurate, segmentation image are not clear. Therefore, thispaper will focus on the pavement crack defect detection, in order to improve theprecision of the detection, from the crack image contrast enhancement, edge detectionand post-processing aspects in-depth research.In the pavement crack image, the cracks information and the background lowcontrast, thus it’s very difficult to detect cracks directly. This paper through the localcontrast enhancement method, combined with mathematical morphology thoughts, andput forward a kind of morphological multi-scale local contrast enhancement method,using the morphological characteristics enhance contrast, effectively improve thefissure image contrast.According to crack image edge detection precision and function of anti-noise ofcoordination problem, the subject based on morphological multi-scale thoughts,according to characteristics of crack’s irregular shape and weak information, putsforward a kind of multi-structural elements anti-noise edge detection operator, andaccording to that different types of crack edge have different filling chances, it sets aalgorithm of self-adaptive weights, which enables the operator test all types of edgecrack, effectively improve the detection accuracy.For the problem of fracture cracks that may be present in test results, Combiningwith the morphology, Different from past based on pixels connection, but adopt theidea of the piece of connection, put forward based on the growth of the active methodof connecting the fracture cracks piece. so as to improve the efficiency and accuracy ofthe connection, make whole segmentation results are clear and complete. |
PDF Full Text Request