Bridge crack image barrier detection and removal method based on generative adversarial network

Abstract

The invention relates to a bridge crack image barrier detection and removal method based on a generative adversarial network. The method comprises the steps that first, multiple barrier pictures are collected, then tags are added, and the pictures with the tags are input into a Faster-RCNN for training; multiple barrier-containing crack pictures are collected, and barrier position calibration is performed through the Faster-RCNN; second, multiple barrier-free crack pictures are collected, and the pictures are turned over to amplify a dataset; third, the amplified dataset is input into the generative adversarial network to train a crack generation model; fourth, information erasure is performed on the positions of barriers in the barrier-containing crack pictures to obtain damaged images; and fifth, the damaged images are input into a cyclic discrimination restoration model for iteration, and then restored crack images are obtained. Through the method, barrier information in the crack pictures can be accurately detected and removed, the peak signal-to-noise ratio of the restored crack images is increased by 0.6-0.9dB compared with before, and therefore a large quantity of crack images with a high restoration degree are generated under a finite crack dataset condition.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a method for detecting and removing obstacles in a bridge crack image based on a generative confrontation network. Background technique [0002] With the development of the transportation industry, road maintenance has become very important. As an important part of the transportation hub in today's society, the bridge not only bears the heavy responsibility of transportation but also concerns the safety of the transportation personnel. However, due to the fact that the bridge structure will inevitably suffer from various damages during long-term use, it will cause the resistance of the bridge structure to attenuate and cause safety hazards. Therefore regular inspection and maintenance is essential. Cracks are the most common defect in bridges. Cracks in bridges can occur for a variety of reasons, mainly due to fatigue of the asphalt pavement, but also unfavo...

AI Technical Summary

Problems solved by technology

The journal ACM proposed the term image restoration in Bertalmio et al.’s Image Inpainting published in PP417–424 in 2000; the journal IEEETransactions on Image Processing published Simultaneous structure and texture image inpainting in 2003 PP882-889 and proposed the use of partial differential Equation method for image restoration, but the algorithm lacks stability, and the restoration results are often poor; then the Mathematical models for local non-texture inpaintings published in 2001 by Chan et al. A unified repair model based on the principle of energy minimization was proposed, but because the model was limited by the size of the repair area and did not satisfy the principle of continuity, it was published in PP436-449 of the Journal of Visual Communication and Image Representation in 2001. In the article Non-texture inpainting by curvature-driven diffusions (CDD), a curvature-driven diffusion model is proposed, but the above algorithms are only suitable for the repair of non-texture images, and there is no way to complete the repair when the image to be repaired is a texture image. Therefore, criminisi et al. published Region filling and object removal by exemplar-based image inpainting in PP1200-121 of the journal IEEE Transactions on Image Processing. Image repair algorithm, which uses blocks as the repair unit and can preserve the texture characteristics of the image, but the algorithm combines the two parts to repair slowly, and cannot repair images containing large continuous area deletions, submitted by Raymond Yeh et al. in 2016 To the conference Computer Vision and Pattern Recognition, the network article Semantic Image Inpainting with Perceptual and Contextual Losses linked to https: / / arxiv.org / abs / 1607.07539 proposes to use deep convolutional confrontational generation network The method of image restoration is based on the method, and the concept of using a binary mask is given to make it possible to restore the texture features and semantics of a damaged image through a well-trained network, but due to the immutability of the binary mask, all Pixels contribute equally to the repaired area in the process of image repair, which makes the repair result often unstable

Images