基于计算机视觉的地面运动观测方法研究

项目名称： 基于计算机视觉的地面运动观测方法研究

项目编号： No.61472373

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 计算机科学学科

项目作者： 戴志军

作者单位： 中国地震局地球物理研究所

项目金额： 80万元

中文摘要： 地震过程中强地面运动的观测是长期以来的一项基础研究工作。目前对强地震动的分析主要依赖于加速计所记录的三方向地震动时间序列。加速度记录在一定程度上能反映地震动强度等方面的信息。但一组三分量加速计只能记录单点运动过程，同时由于地面倾斜等原因使得加速度记录存在基线漂移现象，所以加速计所得记录不能表示大范围多自由度的地面运动过程。 为此，本课题研究基于计算机视觉的强地面运动观测方法。重点解决三个科学问题：加速计记录基线漂移的机理和相应的处理方法、多摄像头多传感器信息融合的方法、多点地面运动的关联关系。然后把方法集成为观测平台并进行实地观测获取数据。通过本课题的研究，可望得到高精度大范围多自由度的地面运动观测方法。因此，本课题的研究具有重要的科学意义和应用价值。

中文关键词： 计算机视觉；特征提取；三维重建；强地面运动；多传感器融合

英文摘要： Observation of strong ground motion under seismic excitation is a long stranding research topic. The analysis on strong ground motion has mainly relied on the three-component time series records of accelerators. The acceleration records could reflect the intensity level of the ground motion. However, a group of three-component accelerators can only record the time-history of a single point. Also, the acceleration records suffer the problem of baseline drift due to a variety of reasons, such as ground inclination, and it cannot represent the ground motion in multi degree of freedom in an extensive area. Thus, we study the observatory method of strong ground motion based on computer vision technology. We will put our main efforts on three scientific problems which are the mechanism of baseline drift and the corresponding processing techniques, the fusion methods of information from multi-cameras and multi-sensors, and the relation among strong ground motion points. Through the research of this project, we are expected to get effective observatory method of ground motion with high precision in multi degree of freedom and in a large extensive area. Therefore, the research of this project has great significance in the academic and important value of practical application.

英文关键词： computer vision;feature extraction;3D recontruction;strong ground motion;multi-sensor fusion