The Algorithm Of Concrete Pavement Crack Image Analysis Based On Parallel Template Matching

Concrete materials have a wide range of applications in road engineering,but cracks are easily generated on the road surface due to their material properties and external factors,which will increase the crack propagation over time and affect the safety of the facility.Therefore,it is necessary to use image processing technology to detect and automate the analysis of pavement cracks,which helps engineers to understand the current situation of cracks in time and use appropriate measures to repair cracks to avoid accidents.In this thesis,the image analysis method of concrete pavement crack based on image processing technology was studied.Firstly,a template matching and high adaptability crack skeleton extraction algorithm was proposed.Then the crack images were accurately classified,and the corresponding feature parameters were selected and calculated for different types of crack images.Finally,a prototype system for crack analysis of concrete pavement was designed and implemented.The main research work of this thesis includes:1.Aiming at the problem that the crack skeleton pixel points extracted by Rosenfeld refinement algorithm are incomplete,the algorithm in this thesis improved the decision template,and refined the judgment strategy combining template matching to solve the phenomenon of partial pixel missing after crack refinement.At the same time,the problem of the non-single pixel width brought about in the improvement process was dealt with.Finally,the initial skeleton of the refined crack was realized without missing points and the skeleton was guaranteed to have a single pixel width,which provided a superior image close to the main feature for the refinement post-processing.2.A high-adaptive burr removal algorithm was proposed to deal with the burr problem still existing in the cracked skeleton.The algorithm in this thesis was different from other algorithms in that the single branch step index was used to judge whether it was a burr,but using the ratio relationship as the judgment criterion,and proposing to compare the ratio of target point in current branch and the entire crack image with the set threshold,thereby effectively removing excess burrs on the refined crack skeleton and being able to adapt efficiently for simple or complex crack images.3.A crack classification algorithm based on projection angle features was proposed in this thesis.The judgment conditions were set according to the different morphological characteristics of cracks,so that each type of crack could be classified.Then,according to the results of crack classification,the appropriate characteristic parameters were selected to calculate and analyze the cracks.Different types of cracks used corresponding characteristic parameters to calculate results,which could more accurately express the damage degree and extension.4.In this thesis,the prototype system of crack image analysis of concrete pavement wss designed,and the crack skeleton extraction algorithm,the crack classification algorithm and the selection and calculation algorithm of crack feature parameters were comprehensively applied.The system correctly processed the input crack image and outputted the correct crack skeleton and corresponding parameter values.