摘要
目前,相似模拟实验中,模型位移及形态等信息完全依赖现场量测,观测效率及容错率较低。针对这个问题,提出一种基于多视影像的相似材料模型三维建模方法。首先,对多视影像生成密集点云的原理方法进行了详细阐述;然后,根据相似材料模型的特点,提出了两种滤波方法对密集点云进行去噪：其一,基于模型标志点深度先验信息滤除密集点云中的大尺度噪声及冗余数据,其二,基于灰度分割的整体局部相结合的K-邻域滤波方法滤除较小尺度噪声;最后,对滤波后的密集点云进行曲面重构、纹理映射等处理,得到直观的数字表面模型。实验结果表明,基于多视影像可以高效地实现对相似材料模型的三维重建;重建模型能够直观准确地表达相似模型的岩层发育形态,为精确量测和分析岩层形变等信息搭建了平台,为矿山压力及开采沉陷等实验研究提供了技术支持。
At present,the information such as the displacement and forms of the model completely depend on field observation,which results in low observation efficiency and low fault tolerance in similar simulation experiments.To solve these problems,this paper presents a method for 3 D modeling of similar material models based on multi-view images.Firstly,it elaborated the method of generating dense point cloud from multi-view images.Secondly,according to the characteristics of similar material model,it presented two filtering methods： one was to filter out the large-scale noise and redundant data in the dense point cloud based on the depth of the mark points,and the other used K-neighborhood method to filter out small-scale noises based on the combination of the whole part and local part of the gray-level segmentation.Finally,it applied surface reconstruction,texture mapping on the filtered dense point cloud and obtained an intuitive digital surface model.The experiment demonstrates that the 3 D reconstruction of similar material models can be achieved efficiently based on multi-view images.The three-dimensional model can intuitively and accurately represent the developing forms of rocks of similar models and build a platform for accurate measurement and analysis of rock deformation and other information,providing technical support for research on mine pressure and mining subsidence.
作者
段志鑫
白志辉
陈冉丽
吴侃
周大伟
DUAN Zhixin;BAI Zhihui;CHEN Ranli;WU Kan;ZHOU Dawei(School of Environment Science and Spatial Surveying,China University of Mining and Technology,Xuzhou Jiangsu 221116,China;Jizhong Energy Fengfeng Group Company Limited,Handan Hebei 056000,China;Hebei Basic Geographic Information Center,Shijiazhuang 050031,China)
出处
《激光杂志》
北大核心
2018年第8期137-141,共5页
Laser Journal
基金
国家自然科学基金青年基金(No.51604266)
江苏省自然科学基金(No.BK20150187)
关键词
相似材料模型
多视影像
密集点云
点云滤波
三维重建
similar material model
multi- view images
dense point cloud
point cloud filtering
threedimensional reconstruction
