成像测井图像纹理特征提取的统计方法研究

纹理特征是利用成像测井图像进行岩性分析的一类重要特征,不同结构和构造的岩石,具有不同的矿物颗粒大小和矿物类型组合及排列形式,在电成像测井图像上就会显示不同的电阻率,反映在电成像图像上就是不同的图像纹理类型。基于灰度直方图、灰度差值直方图、灰度共生矩阵的统计方法,提取相应的纹理特征,通过相似距离法和交会图法对3种统计方法进行了对比分析。不同统计方法在火成岩电成像测井图像上的应用表明,基于灰度差值直方图的统计方法提取的纹理特征,在识别不同结构和构造的火成岩时具有最强的分辨能力,灰度共生矩阵的统计方法次之,灰度直方图的统计方法提取的纹理特征在火成岩的不同结构和构造上不具有差异性。

The Statistical Methods for Extracting Texture Features of Borehole Imaging Logging Diagram

Texture feature was one of the important features of lithologic analysis using borehole imaging logging diagrams.Rocks with different tectonics and structures had different mineral particle sizes,mineral compositions and arrangement forms.Different resistivity and different types of image textures were reflected in electrical borehole images.In this paper,the statistical methods of gray histogram,gray difference histogram,and Gray Lever Co-occurrence Matrix were described.The corresponding texture features from the three statistical methods are compared and analyzed by using similar distances in and cross-plots.The application of different statistical methods in igneous rock logging diagrams shows texture features extracted from the statistical method based on gray level difference histogram have the strongest resolution,it has differentiability in the identification of igneous rock with different structures,and the statistical method based on Gray Lever Co-occurrence Matrix follows.