著者

楠田 啓 西山 孝

出版者

Japan Society of Engineering Geology

雑誌

応用地質 (ISSN:02867737)

巻号頁・発行日

vol.34, no.6, pp.292-299, 1994-02-10 (Released:2010-08-24)

参考文献数

14

被引用文献数

2 2

---

Examination of microcracks by image processing, including pattern recognition and image analysis, is becoming popular in applied geology. In the case of image processing, microcracks in rocks are usually recognized by apparent features, such as differences in brightness, color and/or pattern.On microscopic examination, it is usually difficult to distinguish microcracks from the other elements.The fluorescent method was devised recently to visualize microcracks and cavities in rock and concrete specimens. This method is useful for image processing because images recorded by a CCD camera under ultraviolet ray display a great difference of brightness between microcracks and other texture.We tested five techniques of image processing, linear differential operator (Sobel operator), quadratic differential operator (Laplacian operator;matrix size=3×3 or 9×9), and template matching operator (Kirsch operator and line detection operator), to extract microcracks in weathered granite and andesite fractured by uniaxial compression test retreated by fluorescent method.Actual microcracks in specimens of weathered granite were represented most realistically by thinned binary image obtained by Laplacian operator (matrix size=9×9).Image analysis of microcracks in specimens of andesite fractured by uniaxial compression test was also performed. The results revealed that two types of microcracks, tensile cracks parallel to major principal stress and shear cracks inclined to major principal stress, were extracted.It was also demonstrated that parallel microcracks were formed mainly in groundmass, while inclined microcracks were present mainly in phenocrysts.