The Algorithm Of Concrete Pavement Crack Detection Based On Direction Feature

With the rapid growth of motorway mileage,road traffic has become an indispensable part of people's life.As the most important and most common mode of transportation method in China,road transportation is closely related to people's life.As a motorway surface disease,pavement cracks can easily pose a threat to motorway safety.The pavement crack detection through humans is not only time-consuming,but also easily generates non-uniform detection standard because of humans' subjective judgment so that causes wrong results.Pavement crack detection based on digital image processing technology can overcome the shortcomings of manual detection method,save manpower and cost.It has become a hot spot of current research,and has high application and research value.In this thesis,crack detection algorithms of concrete pavement based on digital image processing technology were studied,moreover,the crack characteristics and difficulties of detection technology were analyzed.Aiming at the defects of the existing crack detection algorithms,improvement and innovation methods were prosposed.The main research works of this thesis are as follows:1.Aiming at the problem of low efficiency of the crack detection algorithm based on percolation model,a high-efficiency crack detection algorithm based on percolation model was proposed by improving the fast percolation algorithm in this thesis.Firstly,the algorithm pre-extracted the cracks of the pavement grayscale image,then refined the pre-extraction result,and finally extracted the refined skeleton points and isolated points,and used them as seed points for percolation treatment to reduce the percolation redundancy points,which improved the efficiency of percolation treatment.2.In order to ensure the recall rate of crack extraction results,a 12-neighbor percolation method was proposed to improve the lack of crack points of detection results in this thesis.Firstly,the skeletal endpoints of the refined pre-extraction result were extracted.Secondly,the 12 neighborhoods of the skeletal endpoints were percolated.Finally,the percolation result was combined with the crack extraction result of the high-efficiency crack detection algorithm based on percolation model of the thesis.So the crack detection algorithm of the thesis can improve the detection efficiency while reducing the lack of crack points.3.Aimed at the fracture phenomenon caused by the percolation process,the directional characteristics of cracks were studied in depth.By improving the crack connection algorithm based on skeleton,a multi-factor decision connection algorithm based on directional feature was proposed.The length of the breakpoint connection,the curvature of the crack tail,the direction factor of the crack end,and the ratio of the dark points of connecting line were combined by this algorithm to find the best connection point.At the same time,according to the ratio of the dark points,the rationality of existence of the connection line was analyzed to eliminate the crack connection lines that should not have existed in the original image.The algorithm of this thesis effectively improved the recall rate of crack detection results,and reduced the phenomenon of misconnection and leakage.4.Combined the high-efficiency crack detection algorithm based on percolation model and the crack connection algorithm based on direction feature,a concrete pavement crack detection prototype system was designed and implemented to achieve efficient and accurate extraction of pavement cracks in this thesis,which facilitated researchers to analyze the parameters of pavement cracks in subsequent steps.