| With the rapid development of highway traffic,the pavement condition detection and maintenance has become a primary task of highway construction.The pattern of cracks is one of the important indicators to measure the pavement quality,so the use of digital ima ge processing technology for pavement crack detection has become a hot spot.In the actual detection process,due to the complexity of the pavement condition,so that there are some disturbance factors such as oil stain,shadows,uneven illumination and ra ndom noise in the collected pavement images.In this case,there will be a miscarriage of justice and cracks undetected with the existing crack detection method,which cannot meet the demand of the detection,and is unable to obtain more accurate and comprehensive crack information,then the pavement maintenance management cannot be carried out timely and effectively.In order to solve the above-mentioned problems,the crack detection in complex pavement images is in-depth studied from the following four aspects in this paper: pavement image enhancement,crack region extraction,breakpoints connection and crack parameter calculation,and crack detection algorithm evaluation.(1)In view of the characteristics of the complex pavement crack images with many interference noises,shadows and uneven illumination,a pavement crack image enhancement algorithm based on Wavelet analysis is proposed in this paper.The low frequency components are enhanced by nonlinear transformation,and the Wavelet threshold for denoising is performed on the high frequency components in the algorithm.The noise information of the high frequency part is suppressed,and then the enhanced image is obtained by wavelet reconstruction.The experimental results show that the algorithm not only enhances the contrast of the pavement image,but also preserves the crack edge detail s while suppressing the noise.(2)In this paper,a crack extraction method based on Legendre moments and fractional integral is proposed for pavement crack images with the blurred background,uneven illumination and shadows.First of all,by using Legendre moments to find the best connected domain similar to the reference image.It is to find the optimal fractional order.Then,an optimal order fractional integral mask is used to process the image,and the gray level of the pixels in the image is reduced.Finally,the histogram of the image processed by fractional integral operator is calculated.The optimal threshold is determined by the histogram in order to extract the crack information.In this method,it makes full use of the properties of the fractional integral,the spatial distribution of pixels is taken into account,and the homogeneity of the image is increased.Not only a large number of noise interference points are removed,but also the crack area is extracted more completely.(3)For the extracted cracks are discontinuous,broken and so on,the crack connection algorithm based on region search is used to connect the breakpoints in this paper.The crack neighborhood search is based on the depth-first principle,and then the breakpoints are removed according to the connectivity principle.Furthermore,the length and width of the connected cracks are measured.The skeleton extraction method and the second-order moment Ferret algorithm are respectively used to measure and analyze the crack length and width.(4)In order to verify the performance of the proposed detection algorithm,the algorithm is evaluated from the three aspects of correctness rate,completeness rate and F-measure.In the experiment,for the pavement crack images with uniform background,low contrast and massive shadow,the proposed algorithm is respectively compared with Otsu thresholding,Canny edge detection,minimum spanning tree,K-means clustering and fuzzy C-means clustering algorithms.The experimental results show that the correctness rate,completeness rate and F-measure of the proposed algorithm are higher.It further proved that the proposed algorithm has the better applicability. |
PDF Full Text Request