Font Size: a A A

The Research Of Pavement Crack Image Detection And Recognition System

Posted on:2015-03-02Degree:MasterType:Thesis
Country:ChinaCandidate:Q DongFull Text:PDF
GTID:2298330422984586Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
The highway will be affected by various vehicle repeated wear and a variety of otherfactors in the use process,Eventually,the road will appear damage which Seriously affectingthe normal operation of highway.The main methods for detection of pavement damage ismanual detection current.This detection method not only spend a lot of manpower andresources, but also the efficiency is low,seriously affect the road testing and maintenanceprocess.The surface crack is the main form of pavement damage,so to develop a set ofautomatic pavement crack detection and recognition system has the important theorysignificance and use value.Based on the digital image processing technology,this paper studies the pavement crackimage detection and recognition system.The main work and conclusions are as follows:1.This paper first analyzes the structure of highway pavement crack detection system and theselection of various modules and working principle,then we design the system hardware andsoftware.The hardware is mainly composed of a detecting device platform, image acquisitionsystem, image processing system, power supply system, lighting system.The software part isthe pavement crack image processing system based on Matlab GUI.And several actualpavement crack detection test, successfully to verify the correctness of the detection systemdesign and hardware.2.According to the characteristics of highway pavement crack image data is enhanced imagesmoothing on the crack,then use the basis of6kinds of edge detection algorithm to extract thefeature of pavement cracks,then it puts forward the8direction Sobel algorithm,and verify the8direction Sobel detection algorithm has obvious effect on improving the quality of pavementcracks.3.In this paper, the use of corrosion, expansion, thinning of these kinds of mathematicalmorphology processing method to deal with the cracks,then use threshold method todistinguish between Linear cracks and transverse crack,and distinguish the transverse cracksand longitudinal cracks by using projection method,Finally, according to the classification of the calculated transverse crack in the external area and length of the transverse crack andlongitudinal crack,after verification of the fracture type judgment accuracy rate can reachabout80%.
Keywords/Search Tags:pavement cracks images, image enhancement, edge detection, morphology, crackclassification
PDF Full Text Request
Related items