桥梁快速建模及病害识别技术应用研究

Research on Application of Bridge Rapid Modeling and Disease Identification Technology

摘要:

利用无人机近景摄影测量技术采集桥梁表观多维度影像数据，再进行多点配准及自动纹理映射解析计算得到精确的纹理三维模型；同时建立深度学习神经网络模型对各种构件和病害图像进行特征提取和训练，开发出桥梁构件划分与自动识别软件；最后在可视化桥梁三维模型的基础上，应用桥梁表观病害识别系统自动识别出构件中的病害信息，识别准确率可以达到75%以上，识别精度可以达到厘米级，从而实现了定量、快速、智能化的检测，大幅提高了桥梁检测评估效率。

Abstract:

The UAV close-range photogrammetry technology is used to collect the bridge apparent multi-dimensional image data. And then the accurate texture 3D model is obtained by multi-point registration and automatic texture mapping analytical calculation. Meanwhile, a deep learning neural network model is established to extract and train the features of various components and disease images, and the software for bridge component division and automatic recognition is developed. Finally, based on the visual three-dimensional model of the bridge, the disease information in the components is automatically recognized by the bridge apparent disease recognition system. The recognition accuracy can reach more than 75%, and the recognition precision can reach centimeter level so as to realize the quantitative, rapid and intelligent detection, which can greatly improve the efficiency of bridge detection and evaluation.