| People are paying more and more attention to the development of road condition detection technology because it is directly related to the traffic safety of daily life.Many city roads are built with grooves to improve skid resistance of road surface,however,the road condition detection technology of grooved pavement is ignored by people for a long time.And crack detection plays a very important role in all road condition detection indicators.The traditional manual detection methods are gradually replaced by automatic detection methods Most road condition detection technologies are based on image processing.Therefore,how to use image processing technology to automatically detect the cracks in the pavement is a key problem in the field of road detection.It is difficult to detect cracks in the image of grooved cement concrete pavement because the width and shape of groove are similar to crack.In order to overcome above difficulties,this thesis studies the denoising method and crack extraction method of grooved cement concrete pavement image.The main research contents and work of the thesis are listed as follows:(1)In order to solve the problem of low accuracy of the cracks due to uneven brightness and low contrast in the image,the local adaptive contrast enhancement algorithm is improved by introducing the contrast of image mean and standard deviation in local and global region.The results show that the improved algorithm not only enhances the image contrast,but also solves the problem of uneven image brightness.(2)In order to overcome the interference caused by random noise,the classical P-M denoising model is improved by introducing shift-invariant shearlet transform(SIST)theory.To avoid image distortion,the diffusion coefficient of P-M model is improved,and the improved model is applied to the SIST of image.The high frequency sub-band image is processed by the new model while the low frequency sub-band image is directly processed by bilateral filtering.The results show that the denoising algorithm proposed in this thesis has better denoising effect.(3)In order to remove the groove from pavement image,a new pavement image smoothing model based on Unidirectional Total Variation(UTV)model is proposed.In order to overcome the shortcomings of traditional groove removal methods,this thesis studies the image variation theory.in order to eliminate the residual isolated noise and avoid image distortion at the same time,the fidelity term and sparse optimization term is introduced,the proposed model is optimized by Alternating Direction Method of Multipliers(ADMM)at last.The results show that the notch removal method proposed in this thesis has achieved good results.(4)A new method for determining crack types is proposed.In order to extract the digital characteristics of cracks and analyze them,we need to determine the types of cracks first.In this thesis,we use the method of the number of connected regions and the distribution density comparison method to distinguish the network crack and linear crack,the direction of linear crack is determined by the projection method and rectangular box method.The results show that the proposed method is reliable. |
PDF Full Text Request