Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Road disease picture enhancement method coupling traditional method and deep convolutional generative adversarial network

A deep convolution, road disease technology, applied in the field of image processing, can solve problems such as unconstrained and uncontrollable, and achieve the effect of strong generalization ability, good model effect, and reduction of labor cost and time loss.

Pending Publication Date: 2021-06-25
BEIJING UNIV OF TECH
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Due to the unconstrained and uncontrollable problems of the original adversarial generation network model, this invention proposes a road disease image enhancement method that couples the traditional method and the deep convolutional adversarial generation network, and uses a convolutional neural network to replace the adversarial generation Multilayer Perceptron in Networks

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Road disease picture enhancement method coupling traditional method and deep convolutional generative adversarial network
  • Road disease picture enhancement method coupling traditional method and deep convolutional generative adversarial network
  • Road disease picture enhancement method coupling traditional method and deep convolutional generative adversarial network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0040] The original pavement lesion image data set used in the present invention is a three-channel grayscale image. The specific implementation steps are as follows:

[0041] (1) Manual marking

[0042] The purpose of human labeling is to classify pavement disease datasets for supervised learning. In supervised learning, the size of the data set and the consistency of classification features will have a great impact on the prediction accuracy of the network. Therefore, in this step, the present invention uses manual calibration method to classify and screen, and labelImg is used to calibrate the pavement disease picture collection, and obtains transverse cracks, longitudinal cracks, reticular cracks, potholes, pavement markings and pavement damage markings. Typical pictures of six categories, such as Figure 4 .

[0043] (2) Batch cutting

[0044] In order not to destroy the classification features such as pavement cracks, but also to reduce the difficulty of training ca...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a road disease picture enhancement method coupling a traditional method and a deep convolutional generative adversarial network. The traditional method is combined with the deep convolutional generative adversarial network to amplify data so as to achieve the effect of data enhancement. A manual method is used for carrying out collection, batch cutting and data set making on an original damaged pavement. According to the disclosed method, based on the generative adversarial network model, the convolutional neural network is used for replacing a multi-layer perceptron in the generative adversarial network, and through mutual game of the generator and a discriminator and continuous iteration updating, the generator has good pavement disease picture generation capability, and the discriminator has good picture authenticity identification capability. According to the road disease picture enhancement method, a new high-quality picture can be generated according to a real pavement disease picture, the purpose of expanding a data set is achieved, good conditions are provided for pavement disease recognition of deep learning, the effect of the trained model is better, and the generalization ability of the model is higher.

Description

technical field [0001] The invention belongs to the field of image processing, and relates to a road disease picture enhancement method coupled with a traditional method and a deep convolutional confrontation generation network. The invention is applied to data enhancement of road disease pictures. Background technique [0002] Cracks, potholes, ruts and other diseases caused by factors such as vehicle load and environment reduce the safety and comfort of driving, and it is necessary to maintain and repair the road surface in time. Due to the lack of sufficient basic theory and technical support, most pavement diseases are currently inspected through on-site inspections. This method requires a lot of time and resources, is inefficient, and cannot meet the needs of modern intelligent pavement inspections. With the development of artificial intelligence technology, machine learning methods are gradually applied to the field of road surface disease identification, which can re...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T5/50G06T3/40G06T3/60
CPCG06T5/50G06T3/4007G06T3/60G06T2207/20081G06T2207/20084G06T2207/20132Y02T10/40
Inventor 陈宁侯越陈艳艳陈逸涵徐子金史宏宇刘卓赵世博
Owner BEIJING UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products