基于图像纹理分析的井下电视岩层裂缝分区方法研究

Study on the Downhole Television Strata Fracture Zoning Method Based on Image Texture

井下电视成像技术能够提供井下直观的图像信息,广泛地应用于工程勘察领域。在井下电视图像资料解释中,通常基于图像显示的地层岩性、产状和构造发育情况,利用人工来推断地层分层分区的边界。为降低人工识别的主观影响、提高识别效率,本文提出基于2D Log-Gabor滤波器的井下电视岩层纹理特征分区方法,通过基于最佳聚类数选择的K-均值聚类算法(Optimal K-means Clustering Selection Algorithm,OKMS算法)实现不同纹理特征区域井下地层的分区。实际应用表明:基于2D Log-Gabor滤波器的井下电视岩层裂缝分区方法能够对岩层裂缝、破碎带和完整岩层进行客观准确的分区,自动分区效率和准确率较人工识别提高了36%,该方法为井下电视岩层裂缝自动分区提供了参考依据。

Downhole television imaging technology can provide intuitive image information underground and is widely used in the field of engineering exploration.In the interpretation of downhole television image data,it is usually based on the lithology,occurrence and structural development of the strata displayed in the image,and manual inference of the boundaries of strata stratification and zoning is used.In order to reduce the subjective impact of manual recognition and improve recognition efficiency,this paper proposes a dounhole television rock texture feature partitioning method based on 2D Log-Gabor filter.By using the optimal K-means clustering selection algorithm(OKMS algorithm)based on optimal clustering number selection,the dounhole strata in different texture feature areas are partitioned.The practical application shows that the downhole television rock fracture zoning method based on 2D Log-Gabor filter can objectively and accurately partition rock fractures,fractured zones and intact rock layers.The efficiency and accuracy of automatic zoning are improved by 36%compared to manual recognition.This method provides a reference basis for the automatic zoning of dounhole television rock fractures.