Research And Development Of Key Technologies For Pavement Crack Detection System

The problem of pavement crack detection can be belonged to the image texture classification.In recent years,the main research direction,in the field of pavement crack detection,has focused on the extraction and calculation of various crack parameter information on the road surface with high efficiency and low loss,especially the relatively small crack in the initial stage.Due to the irregular direction change of the crack on the high-speed road surface,and the complex noise source such as texture,oil spot,object image,rut,marking and other normal asphalt stones,there are still many difficulties in the automatic recognition and classification of pavement cracks.The use of traditional manual detection of pavement crack cycles often reaches 3-6 months,which seriously restricts road maintenance decisions and has lost the timeliness to a certain extent.At the same time,the use of manual pavement crack detection is affected by factors such as insufficient automation of the detection technology and high subjective judgment of the tester,while accuracy and stability are often affected.The thesis studies the methods of classification and identification of crack in high-speed pavement using machine learning technology and image processing technology.The main purpose of machine learning is to extract the area of crack from pavement images.That including crack sample selection,eigenvalue calculation,classifier training and crack circumscribed rectangle calculation.The main steps include crack sample selection,eigenvalue calculation,classifier training,and crack circumscribed rectangle calculation.The optimized Haar template and improved detection mechanism are used to reduce the training time of pavement crack classifier and improve the speed of pavement crack recognition and detection.In the thesis,the crack feature measurement mainly uses crack skeleton and contour information to obtain various parameters such as crack length,width and area.The crack skeleton extraction uses a thinning algorithm.The thinning algorithm is applied to the crack skeleton extraction,and then the burr elimination and skeleton tracking are performed on the refined crack skeleton.In the length calculation,the thesis uses the linear fitting method to calculate the length of crack segment,and then calculate the total length.In the calculation ofthe width,the thesis designs an algorithm which search the edge of the crack contour in the neighborhood circle of the skeleton to realize the crack width calculation.In the calculation of blackness,according to the definition of block degree,a minimum closed skeleton element search algorithm is designed to search for the block value of the calculated mesh crack.In the area calculation,the an area calculation method was designed to calculate the area of the map cracking.Finally,we use OpenCV to realize the function of high-speed road surface crack parameter measurement on Visual Studio 2015 platform.The system designed by the thesis has important application value and economic benefits for shortening the pavement maintenance cycle.At the same time,the system can also be applied to many fields such as bridges and tunnels.