期刊文献+

基于灰度共生矩阵的基岩滑坡纹理特征提取分析研究

Research on Extraction of Bedrock Landslide Texture Feature Basedon Gray Co-occurrence Matrix
下载PDF
导出
摘要 利用纹理特征可以识别滑坡,通过灰度共生矩阵可以很好地获取纹理特征参数,从不同角度对纹理进行刻画。如何合理选取纹理特征参数描述纹理是一个重要的问题。因此,本文采用灰度共生矩阵结合无人机滑坡影像获取纹理特征参数,选取了5个典型的参数(熵、对比度、角二阶矩、逆差矩、相关性)进行分析研究。利用平均梯度、信息熵为指标对五参数进行精度评价。结合东川大白泥和法土村两个典型滑坡为例进行研究,结果表明:东川大白泥滑坡的纹理特征参数中熵、对比度的平均梯度值和信息熵为最高,分别为14.58、6.93和12.24、6.85;相关性的平均梯度和信息熵值为最低,值为6.6、4.8。法土村滑坡的纹理特征参数中熵、对比度的平均梯度和信息熵最高,分别为15.16、7.11和12.24、6.85;相关性的平均梯度值和信息熵值为最低,值为8.5、4.7。经过实地勘察发现东川滑坡面形成大量冲沟和裂缝,纹理结构复杂,信息量丰富;法土村滑坡体与周围地物存在明显差异,纹理基元对比强烈,纹理效果明显。实验结论与实地勘察相符,表明最适合滑坡纹理信息提取的参数是熵、对比度,提取精度高、数值稳定,适合不同类型的滑坡。 Landslides can be identified by using texture features,and the texture feature parameters can be obtained well through the gray level co-occurrence matrix,the texture can be described from different angles.It’s important to select texture feature parameters reasonably to describe texture.Therefore,this paper uses the gray level co-occurrence matrix combined with UAV landslide images to obtain texture feature parameters.We select five typical parameters(Entropy、Contrast、Angular Second Moment、Inverse Difference Moment、Correlation)for analysis and research.Using the average gradient and information entropy as indicators to evaluate the accuracy of the five parameters.Taking two typical landslides of Dong Chuan Dabaini and Fatu Village as examples,the results show that the average gradient value and information entropy value of entropy and contrast among the texture characteristic parameters of Dabaini landslide in Dongchuan are the highest,which are 14.58,6.93 and 12.24,6.85,respectively;The average gradient and information entropy values of the correlation are the lowest,the values are 6.6,4.8.Among the texture feature parameters of Fatu village landslide,the average gradient and information entropy of entropy and contrast are the highest,which are 15.16,7.11 and 12.24,6.85,respectively;The average gradient value and information entropy value of correlation are the lowest,with values of 8.5 and 4.7.After field investigation,it was found that a large number of gullies and cracks were formed on the Dongchuan landslide surface,with complex texture structure and rich information;There were obvious differences between the Fatu landslide and the surrounding features,the contrast of texture primitives was strong,the texture effect was obvious.The experimental conclusion is consistent with the field survey,which shows that the most suitable parameters for landslide texture information extraction are entropy and contrast,which have high extraction accuracy and stable values,and are suitable for different types of landslides.The research results of this paper provide reference and reference value for identifying the texture characteristics of bedrock landslides by using the gray level co-occurrence matrix.
作者 王丽霞 史正涛 喜文飞 李国柱 杨正荣 WANG Lixia;SHI Zhengtao;XI Wenfei;LI Guozhu;YANG Zhengrong(Faculty of Geography,Yunnan Normal University,Kunming 650500,China;Haiju Geographic Information Technology Co.,Ltd.,Kunming 650000,China)
出处 《城市勘测》 2023年第2期187-192,共6页 Urban Geotechnical Investigation & Surveying
基金 云南省科技厅面上项目(202001AT070059) 云南省自然科学基金重点项目(202101AS070019)。
关键词 灰度共生矩阵 特征参数 纹理特征 特征提取 gray co-occurrence matrix feature parameter texture feature feature extraction
  • 相关文献

参考文献24

二级参考文献221

共引文献957

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部