A semi-automated methodology for discontinuity trace detection in digital images of rock mass exposures

Abstract The in situ measurement of discontinuity geometry at rock mass exposures is a slow process, it can be hazardous, and often a large proportion of the exposure is inaccessible. Photogrammetry is a safer method with which to measure discontinuity geometry, and by removing the issues of inaccessibility it provides greater access to the exposure. The use of digital images — rather than customary analogue photographic images — for photogrammetry retains these benefits, while offering the prospect of an automated, and hence fast, method of discontinuity detection and discontinuity geometry analysis. This paper presents a methodology for semi-automated discontinuity trace detection in greyscale digital images of rock mass exposures. The methodology detects discontinuity traces as individual objects, which is a necessary precursor to automated discontinuity geometry analysis. The methodology initially considers a rock mass exposure digital image as a discrete surface, the elevation of which is given by pixel brightness levels. By doing so, a discontinuity trace can be likened to a topographic ravine and therefore some pixels within a discontinuity trace can be found by locating the so-called ‘ravine pixels’. A series of digital image processing techniques are then applied to group and transform these ravine pixels into linear structures that are more suited to computer decision-making, and this results in what we call ravine-line segments. A novel method is presented that links these ravine-line segments together to achieve discontinuity trace detection. The method considers three criteria during the linking process to help ensure acceptable discontinuity detection. These criteria involve the nature of the image search, the angles used to control the shape of a discontinuity trace, and the brightness of the pixels in the rock mass exposure image. Case studies show the discontinuity trace maps that result when the methodology is applied to several rock mass exposure images, and a comparison is made between a discontinuity trace map produced by the methodology and several drawn by hand.