Automatic Pavement Crack Identification System Research Based On Digital Image Processing

Pavement crack as a kind of pavement structural damage is seted great store by the highway administration and the maintenance department.To the gaining of this kind of disease's data message, the traditional visual and manual road crack detection method is not fit to the development of the way.It is labor-consuming,time-consuming,imprecise,dangerous, costly and also it can affect transportation.So the automatic pavement survey is required. Although the techniques of detecting pavement cracks have made great progress,digital image processing algorithms are still under development because of complex pavement images.This thesis is a part of bituminous pavement automatic detection software system.It describes an method of automatic road crack detection,intelligence recognition and chief feature extraction based on digital image processing,and researchs modern computer image processing technology in road surface application of automatic detection. At the same time it enhances the road image by using a fast correct method above all,then processes segment threshold and extracts linear feature,and the last is the identification of the crack from the original gray image,thereby we can obtain the quite satisfactory experimental result.Digital image processing is the emphasis of this thesis research,and it is also core technologies in the crack analysis system,which mainly includes the image intensification, the image division,the image feature extraction and so on.In view of question statement,we research these parts,and compare with several algorithm characteristics.Then,we have used median filtering and OTUS to process the pavement crack image.It basicly can supply the crack information for the characteristic parameter gain.The calculation for crack feature parameters is data foundations of crack evaluation.In feature extraction part,we refine the crack image for further the analysis.Through pixel statistics,calculating the actual area,the length,width and so on characteristic information of the crack.The result of computer image processing shows that it is feasibility,accuracy and effectiveness to the result of actual road scene testing,and the measurement accuracy and testing result are quite satisfactory.We may preliminary realize to the road surface crack image pretreatment,the image division and the characteristic parameter computation. In addition,these algorithms in the research can be applied to analyze the road surface crake automatically,and provide the scientific basic pavement detection data for the highway administration and the maintenance department's decision-making,also be used for reference in other similar image processing cases.