Abstract: To extract accurately rock discontinuities from the borehole images and measure automatically the corresponding geometric information, an intelligent recognition method based on the edge detection algorithm was proposed in this study. Firstly, the feature parameters related to gray level co-occurrence matrix were calculated to locate the rock discontinuities areas in the long borehole images, and Canny edge detector was employed to identify the rock discontinuities edge in these located rock discontinuities areas. Secondly the upper and lower edges of rock discontinuities were extracted through the edge-edge connection and appropriate threshold selection. Thirdly, the rock discontinuities edge trigonometric function was fitted to obtain the sinusoidal curve of the upper and lower edges of the rock discontinuities, and several geometric parameters of the rock discontinuities, such as the dip direction, dip angle and aperture were computed through spatial geometric theory. In the case study, images of borehole wall were recorded in two boreholes in the dam site of Rumei hydropower station, and the geometric data of rock discontinuities derived from the proposed method matched the data obtained from other existing approaches, verifying the feasibility of this method.

Key words: rock discontinuities, borehole image, Canny detection, gray level co-occurrence matrix, orientation, aperture