Road crack segmentation method based on genetic algorithm and U-shaped neural network

Abstract

The invention provides a road crack segmentation method based on a genetic algorithm and a U-shaped neural network. A genetic algorithm is used to search a full convolutional neural network architecture of a U-shaped coding and decoding structure, so as to achieve automatic design. The problems that a road crack segmentation neural network model which is manually designed is tedious in work and large in workload, the designed model is complex, and road crack segmentation is inaccurate under complex conditions are solved, and road cracks can be automatically and accurately segmented. A structure and operation which are better than those of manual design and lower in calculation complexity are found for the modules, high robustness is achieved for interference of complex road surface image cracks, disease non-uniformity and illumination imbalance, the characteristics of the road cracks can be extracted more accurately, and therefore the segmentation accuracy of the whole image is improved.

Description

technical field [0001] The invention belongs to the technical field of structural health monitoring and image processing, and in particular relates to a road crack segmentation method based on a genetic algorithm and a U-shaped neural network. Background technique [0002] With the development of the transportation industry, road maintenance has become very important. Cracks are the most common defects in road damage, and the detection of road pavement defects is the premise of subsequent maintenance and repair. Therefore, the detection of road cracks is essential. In the actual detection process, the distribution of cracks is disorderly and irregular, and it is easy to be interfered by surrounding obstacles, resulting in missed detection and false detection, which poses a great safety hazard to the health of the road. [0003] Traditional road crack identification is generally detected manually by road maintenance personnel on site. Although camera equipment is used for i...

AI Technical Summary

Problems solved by technology

[0014] The purpose of the present invention is to propose the method based on genetic algorithm and U-type neural network road crack segmentation, to solve one or more technical problems existing in the prior art, at least provide a kind of beneficial selection or create condition

[0015] The method of road crack segmentation based on genetic algorithm and U-shaped neural network proposed by the present invention uses genetic algorithm to search the fully convolutional neural network architecture of U-shaped encoding-decoding structure to realize automatic design, and is used to solve artificially designed road cracks Segmenting the neural network model is cumbersome, the workload is heavy, and the designed model is more complicated and the segmentation of road cracks is not accurate in complex situations. The present invention is based on the U-shaped encoding and decoding structure neural architecture. In a specific search In the space, the genetic algorithm is used to search the internal structure of different modules of U-Net to realize the automatic design of the light U-shaped convolutional neural network model. The designed light neural network model can automatically and accurately segment the road cracks

Images