An iterative binary road crack detection method based on UAV image

Abstract：Road crack is an important index to evaluate pavement performance.At present, the commonly used methods include manual detection and road information inspetion vehicle detection, which have the disadvantages of high risk and cost, yet the consumer UAV has huge advantages in getting information about road cracks.There are the shortcomings of large training samples and long test period of the existing road crack detection methods, and a road crack detection method based on UAV image was proposed.A clear binarized image was obtained by an iterative binarization method after performing weighted average grayscale, histogram equalization, and median filtering denoising on the image, and the crack position was identified and marked by integral projection.The results showed that the effect of the iterative binarized road surface crack detection method was significantly improved, compared with the direct detection of the original data and the denoised data based on the gray value feature.When the number of iterations time was 15, the correct rate of recognition of pavement crackscan reached 86.6%, and the misjudgement rate was 13.4%.The robustness was higher in complex scenes, and the efficiency of road crack detection was improved.

Keyword：UAV image;crack detection;iterative binarization;automatic recognition;