基于模式跟踪和路径搜索的岩心CT序列图像裂缝分割  

Crack segmentation of core CT sequence image based on pattern tracking and path search

作　　者：李博[1] 熊淑华[1] 滕奇志[1] LI Bo;XIONG Shuhua;TENG Qizhi(College of Electronics and Information Engineering,Sichuan University,Chengdu Sichuan 610065,China)

机构地区：[1]四川大学电子信息学院,成都610065

出　　处：《计算机应用》2022年第S01期327-332,共6页journal of Computer Applications

基　　金：国家自然科学基金资助项目(62071315)。

摘　　要：岩心计算机断层扫描(CT)序列图裂缝分割是数字岩心裂缝分析的前提。针对传统基于阈值和形态特征的分割方法难以有效对裂缝目标进行批量分割的问题,提出一种基于裂缝位置连续性的裂缝分割方法。首先,利用人工交互对首帧图像进行分割,获得参考模式;其次,对相邻帧进行阈值分割,并将分割结果与参考模式进行与运算和或运算,获得去除噪声的裂缝位置模式;最后,在裂缝位置模式内搜索相邻帧中的裂缝目标连通域,并采用A*算法对断裂的裂缝目标进行连接,形成完整的裂缝目标。对于交叉缝,采用8邻域特征模板提取交叉点,将交叉缝分割简化为单裂缝分割。实验结果表明,所提方法能够在去除噪声的同时对单裂缝和交叉缝进行批量完整分割。Crack segmentation of core Computed Tomography(CT)sequence image is a prerequisite for digital core crack analysis.Aiming at the problem that traditional segmentation methods based on threshold and morphological characteristics are difficult to effectively segment crack targets in batches,a crack segmentation method based on the continuity of crack locations was proposed.Firstly,manual interaction was used to segment the first image to obtain the reference pattern.Secondly,threshold segmentation was performed on the adjacent frame,and the result of the segmentation would be performed AND operation and OR operation with the reference pattern to obtain the crack location pattern with noise removed.Finally,the connected domains of the crack targets in the adjacent frame were searched in the crack location patterns,and the A*algorithm was used to connect crackd crack targets to form complete crack targets.For cross cracks,the 8-neighbor feature was used to extract the intersection points,and the cross crack segmentation was simplified to a single crack segmentation.Experimental results show that the proposed method can perform batch complete segmentation of single cracks and cross cracks while removing noise.

关 键 词：CT序列图 裂缝分割 位置连续性 路径搜索 交叉缝 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]