Intelligent interpretation of digital images of rock joints based on pattern recognition

摘要

As a composition of rockmass, structural planes are one of the chief factors that can largely control the mechanical response of rockmass and its models of deformation and failure under the action of engineering load. Rock joints are a typical sort of structural planes, which are numerous, random and universal in the rockmass. So the measurement and statistics of rock joints are the basics for classifying rockmass grades, evaluating rockmass quality, and selecting physico-mechanical parameters of loc kmass. However, the traditional in situ investigation using manual measuring tools is inefficient and costly. In order to improve the above-mentioned status, new methods to measure joints should be developed, which should be efficient, simple and conveni ent. The rapid developments of modern digital photography and digital image processing make it possible that digital photograph devices such as digital cameras or digital videos can be used to measure joints. So based on theories and methods of pattern r ecognition and image processing, the intelligent interpretation of digital images of rock joints is studied, and the corresponding computational method is also put forward for the first time. The procedure and results of intelligent interpretation of a j oint image have been discussed in detail.