In view of the irregular trace distribution of rock discontinuities, rock mass appears as both a statistical distribution and a texture distribution in the spatial image. This paper proposes a new method on statistica...
详细信息
In view of the irregular trace distribution of rock discontinuities, rock mass appears as both a statistical distribution and a texture distribution in the spatial image. This paper proposes a new method on statistical texture analysis for automated demarcating the homogeneous domains of trace distribution within a rock mass. Grey-Level Co-occurrence Matrix (GLCM) is used to quantify the statistical texture features of trace distribution. Relativity, Inverse Difference Moment and Entropy are screened from ten texture parameters of GLCM using robustness analysis and using principal components analysis. The reliability of three screened texture parameters is verified by comparing the Chebyshev polynomials fitting of three screened texture parameters with Normal distribution, Fisher distribution, and Exponent distribution using chi(2) testing. Automated demarcation of the homogeneous domains is implemented by means of classifying three texture parameters of Relativity, Inverse Difference Moment and Entropy in a moving window using the iterative self-organizing data analysis techniques algorithm (ISOdata). The screening process of texture parameters and a case study indicates that texture parameters and automated demarcation method is so robust, reliable, and efficient that it could replace the traditional representation of the probability statistics in trace distribution and greatly save a lot of manual labor in a large-scale domain.
Hyperspectral image classification is an important part of the hyperspectral remote sensing information processing. The iterativeselforganizingdataanalysistechniquesalgorithm (ISOdata) clustering algorithm which ...
详细信息
ISBN:
(纸本)9781479958368
Hyperspectral image classification is an important part of the hyperspectral remote sensing information processing. The iterativeselforganizingdataanalysistechniquesalgorithm (ISOdata) clustering algorithm which is an unsupervised classification algorithm is considered as an effective measure in the area of processing hyperspectral images. In this paper, an improved ISOdataalgorithm is proposed for hyperspectral images classification. The algorithm takes the maximum and minimum spectrum of the image into consideration and determines the initial cluster center by the stepped construction of spectrum accurately. The classification experiment results show that using the improved ISOdataalgorithm can determine the initial cluster number adaptively. In comparison with the SAM (Spectral Angle Mapper) algorithm and the original ISOdataalgorithm, a better performance of the proposed ISOdata method is shown in the part of results.
暂无评论