ROCK GEOMETRIC FEATURE SURVEY BASED ON CURVELET TRANSFORM AND MIGRATION FOR BOREHOLE GPR IN TONGLUSHAN OLD MINE SITE

摘要

Tonglushan old bronze mine site located at Huangshi city, China has highly historical, cultural and academic value. However, some of the historical remains had suffered from local deformation, surface cracking and other geological disasters recently. Thus structural abnormalities of rock-mass in deep underground were surveyed by borehole Ground Penetrating Radar (GPR) to find out whether there was any mined gallery or mined-out broken area owing to illegal exploitation in recent years. The recorded echo signals suffered from serious interference of the direct waves, multiple scattering waves, clutters, as well as migration problem. Taking these problems into account, one feature extraction scheme based on Curvelet transform combined with migration was proposed. Using the sparsity of GPR signals in Curvelet domain, this method is effective in suppressing the noises with feature information held and extracted. The processed results provide an effective approach for complicated GPR data analysis and interpretations. It is operational to study the underground geological structure and the disaster development about the Tonglushan old mine site.