Research Of Pavement Crack Detection Based On Fractional Calculus

As an early form of the road disease, the pavement crack is a focus in highway maintenance and management, so the global researches have highlighted how to detect the crack efficiently, accurately and intelligently. The development of computer and image processing technologies has accelerated the automation of crack detection, and the pavement crack automatic detection technologies based on digital image processing will contribute to the road conservation and management and further promote the healthy development of the transport services.In this paper, it is focused on the key techniques of the crack detection, such as the pavement image enhancement and the crack extracting, which are to overcome the questions of difficulties of the crack detection in the noisy and complex pavement image. These techniques are combined with digital image processing methods and the effects on the signal from the fractional calculus. The main research content and the contributions of this thesis can be summalized as:1. Research on the basic theory of the fractional calculusThe thesis has researched the derivation of the definition, basic properties, the Laplace and Fourier transforms of the fractional calculus, and focused on the effects on the signal from the fractional calculus. This will establish the theoretical basis for the use of the fractional calculus in pavement crack image enhancement and extraction.2. Image preprocessingAccording to the characteristics of the pavement image, this paper has illustrated the enhancement algorithms in crack image preprocessing. Moreover, a new enhancement algorithm based on fractional differential and minimum means, which integrates the advantages of the fractional differential and the minimum means, has been proposed in this paper. The fractional differential is beneficial in enhancing image texture detail and enhancing the edges and the superiorities of minimum means shows good performance in removing noise and widening crack information. This new enhancement algorithm is characterized with the above advantages.3. Cracks extractionAs a key role in the crack detection, the crack extracting algorithm directly determines the accuracy and precision of the crack detection. As the crack pixel gray is a set of partial minimum to its nearby pixels, this thesis compares the cracks to the valleys in terrain during extracting cracks, and further presents a pavement crack extraction algorithm based on the valley edge detection of fractional integral. Such a new algorithm not only has good anti-noise performance but also achieves satisfactory effects in both improving crack weak borders and accurately positioning crack information in different pavement images.4. Burr noise elimination and crack gaps connectionDue to the fact that the traditional ways don’t have an ideal performance in eliminating burr noise and connecting crack gaps, this paper improves the previous algorithms according to the cracks’features. The improved algorithm is proved to obtain good effects in removing burrs and spurious cracks and can accurately connect crack gaps according to the original crack information.