基于工图图像的法兰三维重建方法研究  被引量:2

Research on Flange 3D Reconstruction Method Based on Engineering Drawings Image

在线阅读下载全文

作  者:于灏[1] 杨建鸣[1] 王小刚[1,2] YU Hao;YANG Jian-ming;WANG Xiao-gang(Institute of Mechanical Engineering,Inner Mongolia University of Science and Technology,Inner Mongolia Baotou014010,China;Coking Plant of Steel Union Corp of Baotou Steel in Inner Mongolia,Inner Mongolia Baotou014010,China)

机构地区:[1]内蒙古科技大学机械工程学院,内蒙古包头014010 [2]包头钢铁钢联股份有限公司焦化厂,内蒙古包头014010

出  处:《机械设计与制造》2020年第11期221-223,227,共4页Machinery Design & Manufacture

基  金:内蒙古自然基金-二维工程图向三维实体的转换研究(2015MS0562)。

摘  要:三维重建课题虽然已经被提出了40余年,但其依然存在效率低、准确性差与模型有歧义等问题。针对上述问题以法兰零件为例提出一种基于工图图像的法兰零件重建方法,方法以工程图纸的扫描图像为研究对象,对其进行降噪、分割与细化等处理,然后采用Harris角点识别算法与Hough圆检测算法提取并统计其上特征,并采用BP神经网络识别并提取零件重建的必要参数,最终实现了法兰模型的三维重建。为工程图纸的三维重建提供了新思路与方法,对后续研究有一定的指导意义。Although the three-dimensional reconstruction project has been proposed for more than 40 years,it still has problems such as low efficiency,poor accuracy and ambiguity in the model.In view of the above problems,this paper proposes a flange part reconstruction method based on the engineering drawings image,taking the flange part as an example.The method takes the scanned image of engineering drawings as the research object,and carries out noise reduction,segmentation and refinement,etc.Then,the Harris corner recognition algorithm and the Hough circle detection algorithm are used to extract and count the features above it,and the BP neural network was used to identify and extract the necessary parameters for the reconstruction of the parts.Finally,the three-dimensional reconstruction of the flange model was realized.It provides new ideas and methods for the three-dimensional reconstruction of engineering drawings,and has a certain guiding significance for follow-up research.

关 键 词:三维重建 工程图纸 图像识别 图像分割 特征提取 BP神经网络 

分 类 号:TH16[机械工程—机械制造及自动化] TP391.4[自动化与计算机技术—计算机应用技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象