运用GMS算法自动识别相似模拟材料实验裂隙场  

Automatic Identification of Experimental Fracture Fields of Similar Simulated Materials by GMS Algorithm

作　　者：李平超 LI Pingchao(Mapping and Geographic Information Institute,Henan Provincial Bureau of Geological and Mineral Exploration and Development,Zhengzhou 450006,China)

机构地区：[1]河南省地质矿产勘查开发局测绘地理信息院,河南郑州450006

出　　处：《地理空间信息》2024年第3期51-54,共4页Geospatial Information

基　　金：河南省科技攻关资助项目(212102310404)。

摘　　要：通常采用沙盘物理模型与近距离摄影测量相结合的方法来探索上覆地层的断裂机制,但由于沙土物理模型裂缝图像的信噪比较低,该算法往往提取得不完整。鉴于此,提出了一种基于计算机视觉的图像梯度原理、图像匹配和区域筛选技术(GMS算法)的裂隙检测算法。首先利用峰峰市九龙煤矿15235工作面的钻孔柱状图建立沙盘物理模型;再利用定焦数码相机拍摄该模型的正射影像;最后根据模型图像的纹理特征,分别采用图像匹配技术、区域筛选技术分割规则的大面积非裂缝纹理和微小非裂缝纹理。研究结果表明,GMS算法具有很强的抗干扰能力,能准确识别并自动提取相似模拟材料中上覆地层的裂隙场。The fracture mechanism of overburden strata is usually explored by the combination method of sand-physical model and close-range photogrammetry.Due to the low signal-to-noise ratio of fracture images from the sand-physical model,the application of this algorithm often leads to incomplete extraction.To solve this challenge,we proposed a fracture detection algorithm based on image gradient principle of computer vision,image matching and region screening techniques(GMS algorithm).Firstly,we used a synthetic columnar section of the 15235 working face of Jiulong coal mine in Fengfeng City to establish a sand-physical model.Then,we took the orthophoto image of model with Nikon digital camera.Finally,based on the texture characteristics of model images,we used image matching technique and area screening technique to segment the regular large non-fractured texture and the tiny non-fractured texture.The results show that the GMS algorithm is highly resistant to interference and can accurately identify and automatically extract the fracture fields of overburden strata in similar simulated materials.

关 键 词：采矿沉降 沙盘物理模型 裂缝检测 计算机视觉 图像处理 

分 类 号：P234.1[天文地球—摄影测量与遥感]