Pavement crack detection method

Abstract

The invention discloses a pavement crack detection method which comprises the following steps: S1, image acquisition: acquiring a road image shot in an actual scene; s2, image processing: preprocessing the road image through image processing operation; s3, model construction: constructing a pavement detection model for pavement crack detection; and S4, pavement crack detection: inputting the processed road image into a pavement detection model, and completing pavement crack detection of the road by the pavement detection model. According to the invention, automatic pavement defect detection isrealized, so that a manager or a driver can quickly and timely master pavement information of a road, and a practical and reliable reference basis is provided for subsequent pavement maintenance.

Description

technical field [0001] The invention relates to a detection method, in particular to a deep learning-based pavement crack detection method, which belongs to the technical field of artificial intelligence. Background technique [0002] With the development of urbanization all over the world, the construction of municipal roads is gradually increasing. Both at home and abroad are researching and exploring how to improve the quality of road construction and the maintenance methods for roads after construction is completed. [0003] Cracks on the road surface are a relatively common phenomenon. In order to ensure the long-term normal use of the road, municipal personnel are required not only to improve the construction quality during road construction, but also to carry out regular maintenance and testing on the road surface. Most of the traditional road inspection work needs to be completed by manual inspection. In the process of road inspection, an inspector can only complete ...

AI Technical Summary

Problems solved by technology

[0006] Through the above-mentioned combing of the domestic pavement crack detection system, it can be seen that although the current road crack detection system has improved the efficiency of manual detection to a certain extent, it also has some serious defects.

First of all, the equipment used in some systems is expensive and expensive to maintain; secondly, the image processing of these systems almost all uses the traditional image recognition algorithm, that is, the method of manual feature extraction, and has not yet adopted deep learning. To realize automatic feature extraction; thirdly, the handheld detection equipment used in some systems is difficult to apply in practice, the use process is time-consuming, laborious and inefficient, and the inspection results are highly subjective, there is no uniform standard in the judgment process, and there are large gaps risk

Images