| 基于红外图像识别技术的道路与桥梁故障诊断 |
投稿时间：2021-08-08 修订日期：2021-11-29  点此下载全文 |
| 引用本文： |
| 摘要点击次数: 97 |
| 全文下载次数: 0 |
|
|
| 中文摘要:针对传统道路与桥梁故障诊断方法识别效率低、数据误差大等问题,提出了一种基于红外线图像识别的智能监控系统,并采用装有红外对射光栅的故障监控机进行监测,实现图像识别、数据管理和应用为一体化。该系统包含分层图像融合框架,采用逐层深度学习技术挖掘图像的细节信息,提取图像的关键信息进行词典学习。根据形态相似性,将源图像分为平滑、随机和主方向的面片。分别采用基于Max-L1和L2-范数的加权平均融合规则对三个图像块组的高频分量和低频分量进行融合。将融合后的低频分量和高频分量进行组合,得到最终的融合结果。对比实验验证了所提出的图像融合方案的实用性和可靠性,在相同的图像分割参数下,本文模型计算得到的故障监测率为94.14%。 |
| 中文关键词:红外对射光栅 图像识别 词典学习 分层图像融合框架 高低频分量 |
| |
| Road and bridge fault diagnosis based on infrared image recognition technology |
|
|
| Abstract:Aiming at the problems of low recognition efficiency and large data error of traditional road and bridge fault diagnosis methods, an intelligent monitoring system based on infrared image recognition is proposed, and the fault monitoring machine equipped with infrared reflection grating is used for monitoring, so as to realize the integration of image recognition, data management and application. The system includes a hierarchical image fusion framework, uses layer by layer deep learning technology to mine the details of the image, and extracts the key information of the image for dictionary learning. According to the morphological similarity, the source image is divided into smooth, random and main direction patches. The weighted average fusion rules based on max-l1 and L2 norm are used to fuse the high-frequency and low-frequency components of the three cluster image block groups. The fused low-frequency component and high-frequency component are combined to obtain the final fusion result. Comparative experiments verify the practicability and reliability of the proposed image fusion scheme. Under the same image segmentation parameters, the fault monitoring rate calculated by this model is 94.14%. |
| keywords:Infrared reflection grating Image recognition Dictionary learning Hierarchical image fusion framework High and low frequency components |
| 查看全文 查看/发表评论 下载pdf阅读器 |
|
|
|
次