全景钻孔图像中结构面全自动识别方法研究

发布时间：2018-03-24 21:11

本文选题：岩石力学　切入点：钻孔图像　出处：《岩石力学与工程学报》2017年08期

【摘要】：数字全景钻孔摄像技术在实际工程上得到了广泛的应用,获得大量的高精度实测钻孔图像。这些钻孔图像准确记录了地质特征信息,特别是结构面的几何形态特征。然而,这些特征信息的获取通常是由人工完成的,工作量大,人为因素较多。针对该问题,提出一种针对全景钻孔图像的结构面全自动识别方法。该方法首先提出用图像灰度梯度合成信号在钻孔深度方向的投影的方法来划分结构面所在区域及其产状范围;然后提出用标准模板正弦函数迭代匹配结构面特征量的方法来搜寻区域内结构面的所有可能正弦曲线,并从中筛选出最优的正弦曲线作为结构面的特征曲线;最后结合区域划分和模板函数匹配的结果,分析特征曲线参数,并进行再匹配和转换,最终得到工程所需的结构面位置、倾向、倾角和隙宽等参数,从而实现钻孔图像结构面的全自动识别。结果表明该方法能够连续快速地全自动识别整个钻孔图像内的结构面,并获得对应的高精度几何参数,算法稳定可靠、结构面识别率高。与传统方法相比,该方法首次实现了整个钻孔图像中结构面的全自动识别与几何参数提取,极大地提高了工作效率,为钻孔图像的后期处理与信息获取提供一种实际可行的有效方法。
[Abstract]:Digital panoramic borehole camera technology has been widely used in practical engineering, and a large number of high-precision borehole images have been obtained. These borehole images accurately record the geological characteristics, especially the geometric features of the structural plane. The acquisition of these feature information is usually done by hand, the workload is heavy, and there are many human factors. A fully automatic recognition method for the structure plane of panoramic borehole image is proposed. Firstly, the projection method of image gray gradient composite signal in the depth direction of borehole is proposed to divide the region of structure plane and its occurrence range. Then, a method of iteratively matching the eigenvalues of structural surfaces using standard template sinusoidal function is proposed to search all possible sine curves of structural surfaces in the region, and the optimal sine curves are selected as the characteristic curves of structural surfaces. Finally, combining the results of region partition and template function matching, the parameters of characteristic curve are analyzed, and the parameters of structural plane position, inclination, dip angle and gap width are obtained by re-matching and converting. The results show that the method can recognize the whole borehole image automatically and continuously, and obtain the corresponding high precision geometric parameters. The algorithm is stable and reliable. Compared with the traditional method, this method has realized the automatic recognition and geometric parameter extraction of the whole borehole image for the first time, and greatly improved the working efficiency. It provides a practical and effective method for the post processing and information acquisition of borehole images.
【作者单位】：中国科学院武汉岩土力学研究所岩土力学与工程国家重点实验室;武汉大学电子信息学院;
【基金】：国家自然科学基金资助项目(41402278,41372317)~~
【分类号】：TU45

【相似文献】