基于改进的Hessian和Live-wire算法的岩石节理裂隙检测  

Rock Fracture Tracing Based on Hessian and Live-wire Algorithms

作　　者：王艳[1] 李晗[2] 陈佳悦 陈卫卫[4] 王梦菲 闫迪 李宏霞 王卫星[5] WANG Yan;LI Han;CHEN Jiayue;CHEN Weiwei;WANG Mengfei;YAN Di;LI Hongxia;WANG Weixing(School of Artificial Intelligence,Xi′an Aeronautical Polytechnic Institute,Xi′an 710089,China;Department of Automotive Engineering,Liaoning Provincial College of Communications,Shenyang 110122,China;Department of Computer Science,University College London,London,UK WC1E 6BT;School of Automation Engineering,Xi′an Aeronautical Polytechnic Institute,Xi′an 710089,China;School of Information Engineering,Chang′an University,Xi′an 710064,China)

机构地区：[1]西安航空职业技术学院人工智能学院,陕西西安710089 [2]辽宁省交通高等专科学校汽车工程系,辽宁沈阳110122 [3]伦敦大学学院计算机科学系,伦敦WC1E 6BT [4]西安航空职业技术学院自动化工程学院,陕西西安710089 [5]长安大学信息工程学院,陕西西安710064

出　　处：《金属矿山》2023年第8期265-271,共7页Metal Mine

基　　金：西安航空职业技术学院2022年度校级教育教学改革研究项目(编号:22XHJG14,22XHJG19)。

摘　　要：在图像处理中,由于岩石节理裂隙是最复杂的线状目标之一,针对该对象的检测算法的研究一直是一个难题。故此,研究了一种新的跟踪裂隙边缘线的算法。首先,若图像尺寸太大,进行有选择的图像缩小,然后采用基于Geodesic Shadow Removal的算法进行图像平滑,再用一种基于Hessian矩阵算法增强模糊及微细节理裂隙;用一种基于Live-wire Contour思想的节理裂隙边缘线特征点提取的算法进行边缘特征点提取;最后基于特征点之间的距离及相应线段夹角来连接特征点以形成完整的线段。选择了200幅图像进行实验,通过与十多种传统和新近的算法相比,新算法能够在复杂的岩石节理裂隙图像中,准确快速地提取节理裂隙边缘线,为在该领域引进深度学习等方法奠定基础。Since rock fracture is one of the hardest linear objects in image processing,the detection algorithm study is hard.In order to quickly and accurately measure and analyze the development of rock fractures,a new algorithm for tracking fracture edges is studied,firstly,if the image size is too large,the image can be shrunk in a certain rules,then,the image is smoothed by the algorithm based on geodesic shadow removal,and the fuzzy and micro fractures are enhanced by a Hessian matrix algorithm;and then it uses the live wire contour based on algorithm extracts edge feature points;and finally according to the distance and angle between neighboring feature points,fracture points are connected to form fracture boundary segments.In the experiments,200 fracture images were selected,by compared with more than ten traditional and recent image segmentation algorithms,the new algorithm can accurately and quickly extract the edges of fractures in complex rock fracture images,and it can be the basis for introducing semantic segmentation methods such as deep learning.

关 键 词：岩石节理裂隙 Live-wire Geodesic shadow HESSIAN 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]