Research On Pavement Crack Recognition Based On Improved Gray Scale Segmentation Algorithm

Pavement crack on the road is one of common diseases, so it attracts more and moreattention from the highway management and conservation departments. With thedevelopment of highway construction, towards obtaining of highway pavement crack’s dataand information, the crack detection method based on manual labor can’t satisfy the need ofactual application gradually, because it consumes a lot of manpower，material resources and itis dangerous, meanwhile it is not accurate even affects people’s travel. Even pavement crackidentification based on image processing technology in some developed countries has certaindevelopment, but because of the complication of the collected image, in order to identifycracks better, for it the digital image processing algorithms become the focus of scholars fromvarious countries.This paper focuses on image processing algorithms for specified image of crack.According to the characteristics of the image, it uses a fast image correction method foruneven gray scale image, enhances quality of image and removes the noise, by studyingdifferent algorithms and comparing several features of various algorithms, this paper attemptsto use the method of the combination of median filtering and mean filtering, adjusts theimage based on the overall average gray of the image, making it easy for image thresholdsegmentation, then the crack of the image is extracted initially, next operate the morphologyto remove the noise and identify the target more clearly. The paper uses the projection methodto classify the cracks, the magnitude of the eigenvalue represents the pavement quality isgood or bad, after the classification of cracks is completed, then calculate the eigenvalue ofcrack by doing statistics to the pixels of different images, the eigenvalue here includinglength, width, area and other data. After a series of processing, compare the results with theresults of manual detection, then it shows that the result of the algorithms used by this paperis accurate and feasible.The system can also perform batch processing for multiple images, so it can improvedetection efficiency of the road with cracks so that it can detect automatically. Further theworkload of the maintenance department and highway administration can be reduced, so theycan make better decisions in terms of road detection, at the same time this paper is a reference to other task needing image processing operations.