Research On Detection And Recognition Technology Of Driveway Pavement Cracks

Along with the use of the driveway,a lot of the early roads appeared damage,deflection and other issues,seriously affecting the driveway carrying capacity,driving safety and service life.In the detection,quantitative indicators of the highway can be used to estimate the expected working life and maintenance workload,the measure of pavement crack is the most important index.with the rapid development of computer technology and the rapid improvement of image processing technology,the automatic detection technology of pavement disease based on image processing is proposed to apply to the actual.In the actual pavement crack detection,the direction of the crack itself is more complicated,in addition,there will be oil,obstacles shadow on the road,this will lead to serious image noise,the realization of ideal detection algorithm is the difficulty in the detection of pavement cracks.At present,the image processing algorithm can not satisfy the actual demand of the crack image processing.in order to detect pavement cracks effectively,the paper makes a further study on the pavement crack image preprocessing,image segmentation,classification of cracks and other aspects.(1)Pavement crack image preprocessing.according to the characteristics of pavement crack image proposed an improved median filtering algorithm and an enhancement algorithm based on region contrast,the pavement crack image can be denoised and enhanced.Compared with commonly used denoising and enhancement algorithm experimental results,the algorithm in this paper can suppress the noise and enhance the crack area,and the image quality is improved obviously.(2)Pavement crack image segmentation.this paper proposed an image difference method based on block in the pavement crack image segmentation,the background difference image is obtained after the extremum smoothing of the phase shift,the binarization after difference,the edge of the crack is clear,and crack edge continuity is better.then,According to the linear characteristics of the cracks,the geometric features of the connected domain are used to further remove the noise.(3)Crack recognition and classification.The projection characteristics and the number of different types cracks are obvious differences,the characteristics of cracks are extracted,using BP neural network to classify the crack image,training the network with different number of hidden layers,choose the right network to classify the actual pavement crack image.finally,the crack skeleton is combined with the geometrical characteristics of the crack,can calculate the parameters of the crack,and calculate pavement damage index PCI to make a comprehensive quantitative evaluation of pavement damage.