Image-based discontinuity identification: Bildgestützte Trennflächenidentifikation

Knowledge about the joint network is important in rock engineering. It controls the stability and strength of blocky rock masses. Thus, the joint network is mapped in more or less detail, regarding the specifications. To do so, different methods have evolved over the years. One of these methods is digital mapping, based on the visual and spatial appearance of the distinct joints. This mapping usually is done by an experienced geologist. However, the trend in digital mapping is towards a supervised but mainly autonomous identification of discontinuities. One aspect of this automated process is the recognition of joints in digital outcrop images by their optical attributes. This contribution presents a density based joint plane detection. By the combination of a vector based joint plane detection and a pixel based joint trace detection, not only joints, but also the foliation in the investigated tunnel face could be extracted very accurately. The approach represents the current state of research and further improvements are to be expected.