| 基于卷积神经网络的路面裂缝检测 |
| |
| 引用本文： | 赵珊珊,何宁. 基于卷积神经网络的路面裂缝检测[J]. 传感器与微系统, 2017, 36(11). DOI: 10.13873/J.1000-9787(2017)11-0135-04 |
| |
| 作者姓名： | 赵珊珊 何宁 |
| |
| 作者单位： | 1. 北京联合大学北京市信息服务工程重点实验室,北京,100101;2. 北京联合大学智慧城市学院,北京,100101 |
| |
| 基金项目： | 国家自然科学基金资助项目 |
| |
| 摘 要： | 对路面图像块预标记,根据预标记结果对路面图像进行强度归一化预处理,在保留裂缝信息的同时,减少背景光照不均的影响.将预处理后的路面图像输入卷积神经网络(CNN)模型实现路面图像裂缝的检测.由于路面裂缝分布复杂,在训练网络时,使用不同尺度和不同角度的路面图像进行模型训练,使得网络能够检测不同裂缝形状.实验结果显示:裂缝检测结果较好.
|
| 关 键 词： | 图像处理 公路裂缝 卷积神经网络 检测 |
Road surface crack detection based on CNN |
| |
| ZHAO Shan-shan,HE Ning. Road surface crack detection based on CNN[J]. Transducer and Microsystem Technology, 2017, 36(11). DOI: 10.13873/J.1000-9787(2017)11-0135-04 |
| |
| Authors: | ZHAO Shan-shan HE Ning |
| |
| Abstract: | Road surface crack detection method based on convolutional neural network (CNN )is proposed. Intensity normalization preprocessing on road surface image according to pre-labeled result,which can reduce the influence caused by the un-illumination background and the white pixel. The CNN is used to realize the crack detection. Use different scales and angles of road image for model training to improve the ability to detect small and scattered road crack. The experimental result shows that this method is effective. |
| |
| Keywords: | image processing highway crack convolutional neural network(CNN) detection |
| 本文献已被 万方数据 等数据库收录！ |
|
