项目名称: 基于多目视觉的复杂管路数字化快速检测新方法研究
项目编号: No.51305031
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 机械、仪表工业
项目作者: 刘少丽
作者单位: 北京理工大学
项目金额: 25万元
中文摘要: 管路实物检测是保证管路制造质量和实现无应力装配目标的不可缺少的最有效手段。随着复杂机电产品的小型化、轻量化方向发展,管路的空间形状越来越复杂,传统的管路检测方法已经无法满足工程需要。本项目拟提出一种基于多目视觉的大尺寸复杂管路快速检测新方法,系统探索该方法中的检测与三维重建相关理论与方法,开展多目数据采集、拼接与冗余数据处理、多目标定、特征提取、立体匹配等技术研究,突破基于测量域的多视点优化布局、基于空间曲线的多视点融合、基于中心线立体匹配的管路三维重建、基于CAD三维模型的管路接头和连通的识别重建、以及测量误差分析与建模技术等关键实现技术。最后研制原型系统并开展应用验证,实现复杂管路的精确测量与三维显示,为最终实现管路一次装配成功和无应力装配提供新的检测方法和手段。
中文关键词: 机器视觉;管路;三维重建;特征识别;在线检测
英文摘要: The detection of pipelines is the most effective means to ensure the pipeline's manufacturing quality and to realize the stress-free assembly target.Along with the complex electromechanical products becoming smaller and lighter, the spatial shape of the pipelines is more and more complex. Then, the traditional detection methods of pipelines have been unable to meet the needs of engineering. This project proposes a new method based on multi-vision for large-size complex pipeline's rapid detection.The theory and methods for detection and three-dimensional reconstruction are researched in this project. The multi-mesh data acquisition, stitching and redundant data processing, calibrating, feature extraction, stereo matching technology will be researched.The key technologies will be break, such as the optimization of multi-view layout based on measurment domain, the pipeling's three-dimensional reconstruction based on the centerline's stereo matching, the pipeline's joints and connectivity identification and reconstruction based on CAD model, and the analysis and modeling techniques for measurment error. Finally,the prototype system will be developed and the verification of application will be carried out.The complex pipelines' accurate measurement and 3D display will be realized. This project will provide a new
英文关键词: machine-vision;tube;3D reconstruction;feature recognition;online inspection