摄影测量计算机视觉在矿山相似材料模拟实验中的应用研究

发布时间：2018-06-26 18:09

本文选题：相似模拟实验 + 摄影测量　；参考：《西安科技大学》2015年硕士论文

【摘要】：矿山相似材料模拟实验变形监测是大型采矿工程模拟实验必做的一项内容,这项内容的主要目的是为实际开采提供安全指导。为了实现变形监测,需要在模型上人为布设大量监测点,通过量测每个监测点在各个工作面上的物方坐标,得到上覆岩层在开采时的变形特征。从岩层的运动规律可以确定出顶板周期来压的活动性质。针对目前量测监测点物方坐标时采用方法存在的局限性,本文提出一种基于摄影测量计算机视觉单应矩阵的三维重建算法,实现了相似材料模拟实验的变形监测。这种算法大致分为以下三步:⑴相机检校。相机检校的主要目的是为了计算描述物方点坐标与其在二维影像上对应像点间几何模型的参数。通过计算结果既可以得到物方与像方之间的投影关系,也可以解决由于普通数码相机引起的影像边缘变形问题,即实现影像预处理。通过对比分析常用的多种检校算法的优缺点,最终选择张正友平面检校来完成检校工作。⑵像点坐标高精度量测。像点坐标量测算法的效率和精度,是本文三维重建算法的关键之一。以回光反射材料制作的圆形标志和编码标志分别作为监测点和编码点,结合Canny算子和椭圆拟合算法实现标志点像点坐标的高精度量测。⑶监测点物方坐标解算。对比分析传统三维重建算法的不足之处,再结合本次实验研究对象为平面模型,以单应矩阵为基础实现三维重建。这种算法可以基于单张影像完成,从而避免复杂的同名点匹配过程。三维重建算法的理论模型经确定后,选择适当的计算机语言来加以实现,本文主要利用MATLAB和C++两种语言。以东峡煤矿建立的物理模型为研究对象,经该算法计算得到监测点物方坐标,在精度满足要求的情况下,利用这一结果绘制每个工作面的变形曲线。将变形曲线和利用高精度测量手段量测到的结果所绘制的曲线进行比较,初步判断该算法是否合理可行;在两种曲线一致的情况下,与物理模型的实际变形作对比,查看得到的曲线是否符合实际情况。通过实例证明了本文提出算法的可行性、计算速度和计算精度,实验结果表明:本文算法实现了全自动监测点物方坐标解算,计算速度快,结果精度高,可以满足相似材料模拟实验。
[Abstract]:Deformation monitoring of similar mine material simulation experiment is a necessary part of large-scale mining engineering simulation experiment. The main purpose of this content is to provide safety guidance for actual mining. In order to realize deformation monitoring, it is necessary to set up a large number of monitoring points artificially on the model, measure the coordinates of each monitoring point on each face, and obtain the deformation characteristics of overlying strata in mining. According to the movement law of rock strata, the active properties of roof periodic pressure can be determined. In view of the limitation of the method used in measuring the coordinates of objects at present, a 3D reconstruction algorithm based on the monoclinic matrix of photogrammetry computer vision is proposed to realize the deformation monitoring of similar material simulation experiments. This algorithm is roughly divided into three steps: 1 camera calibration. The main purpose of camera calibration is to calculate the parameters of the geometric model between the objects' square points and their corresponding points in two-dimensional images. The projection relation between object side and image square can be obtained by calculating results, and the image edge deformation problem caused by ordinary digital camera can also be solved, that is, image preprocessing can be realized. By comparing and analyzing the advantages and disadvantages of many commonly used calibration algorithms, Zhang Zhengyou plane calibration was selected to complete the high precision measurement of the image coordinates of the calibration work .2. The efficiency and precision of the image point coordinate measurement algorithm is one of the keys of the 3D reconstruction algorithm in this paper. The circular mark and the coded mark made by the reflective material are used as the monitoring points and the coding points, respectively, and the Canny operator and the ellipse fitting algorithm are combined to calculate the square coordinates of the image points of the mark points. By comparing and analyzing the shortcomings of traditional 3D reconstruction algorithm, and combining the object of this experiment as plane model, the 3D reconstruction is realized on the basis of monoclinic matrix. This algorithm can be completed on the basis of single image, thus avoiding the complicated matching process of points of the same name. After the theoretical model of 3D reconstruction algorithm is determined, appropriate computer language is chosen to realize it. This paper mainly uses MATLAB and C language. The physical model established in Dongxia Coal Mine was used as the research object. The coordinates of the monitoring points were calculated by the algorithm. The deformation curve of each working face was drawn with this result under the condition that the precision satisfied the requirement. By comparing the deformation curve with the curve drawn by using the high precision measurement method, the paper preliminarily determines whether the algorithm is reasonable and feasible, and makes a comparison with the actual deformation of the physical model when the two curves are the same. See if the resulting curve matches the actual situation. The feasibility, calculation speed and accuracy of the proposed algorithm are proved by an example. The experimental results show that the algorithm has realized the calculation of the square coordinates of the fully automatic monitoring points, and the calculation speed is fast and the precision of the results is high. It can satisfy the simulation experiment of similar materials.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TD326

【相似文献】