Pavement defect damage real-time monitoring method and system

Abstract

The invention discloses a pavement defect damage real-time monitoring method. The method comprises the following steps: acquiring road image data; preprocessing the road image data to obtain a pavement defect data set; enhancing the collected pavement defect data set; building a pavement monitoring network framework, and generating a pavement defect monitoring model; training the pavement defect monitoring model to generate a trained pavement defect monitoring model; carrying out shooting in the vehicle driving process, analyzing the pavement defects in real time through the trained pavement defect monitoring model, and pushing an analysis result. The invention further discloses a system based on the pavement defect damage real-time monitoring method. According to the invention, by enhancing the utilization of low-level features and the advantages of a full-connection network, each evaluation index of the algorithm is improved; the influence of an interference object on pavement defect identification can be reduced, the pavement crack identification effect is improved, the pavement state is rapidly evaluated, and the design of a highway construction scheme is facilitated.

Description

technical field [0001] The invention belongs to the field of pavement defect monitoring, in particular to a method and system for real-time monitoring of pavement defect damage. Background technique [0002] The quality of highway pavement has always been an important issue in the industry's in-depth research in highway construction schemes. At the same time, after a long period of wear and tear on the road, the geological structure has changed, resulting in problems such as surface texture scattering, cracking, and potholes on the road surface. First, before building a pavement crack detection model, it is necessary to characterize the collected pavement crack data set Classification, by classifying pavement cracks, can be of great help in building models and experiments. According to the degree of damage to the pavement, geometric characteristics, texture, and comprehensive factors, pavement defects are mainly divided into five categories: longitudinal cracks, transverse ...

AI Technical Summary

Problems solved by technology

[0003] However, the quality of these existing pictures is not good. Apart from cracks in some pictures, there are also some interfering objects such as road shoulder fences, road marking lines, side cars and road leaves, which will interfere with the identification of road defects. ,like figure 2 Shown is a schematic diagram of other disturbance factors besides cracks and shade, figure 2 a means a shoulder fence, figure 2 b means fallen leaves, figure 2 c means sidecar, figure 2 d represents the identification line

For such complex interfering road surfaces, the detection effect of traditional image detection algorithms will be greatly reduced, the robustness is not strong, and it is easy to misjudge non-cracks in the picture

like image 3 a-3d is a schematic diagram of the interference of tree shade on road surface recognition. Due to the angle of shooting and the shadow caused by sunlight, there is no obvious difference in the contrast between the crack and the surrounding area, which will also interfere with the detection of road cracks.

Images