Surface Reconstruction For Oil Exploration

In oil exploration, geological data are obtained through different exploration means. The complex geological model is then expressed from these data, which will provide a better observation for geologists, and assist geologists to discover oil-gas resources. This thesis studies and analyses the current surface reconstruction technology, aims at different sinarios which may occur in the oil exploration process. It presents different reconstruction methods for different data organizations, including volume data and parallel cross sections. Firstly, traditional surface reconstructions from volume data are usually designed for a single surface, which do not have a high requirement for efficiency. However, geological reconstruction algorithms need to deal with a large scale of volume data of multi surfaces as well as geological constraints between surfaces. As a result, they are hard to achieve a real-time rendering speed. We take advantage of the GPU parallel computing ability and design a parallel algorithm for reconstructing multi-geological surfaces in real time. Our method can reconstruct multi-surfaces under geological constraints without using Boolean operations on solids. Secondly, for the surface reconstruction from parallel contours, we use a function blending to generate the transition surface and employ Kochanek-Bartels splines to provide a local editing feature near contours which will make the model more resemble the real geology. Finally, we present a modeling method based on skeletons and contours designed for the reconstruction of tubular topography. By utilizing rotation minimizing frames, we convert this problem into normal surface reconstructions from parallel contours. A tubular surface can be constructed from simple sketches by our method. When editing the skeleton, the surface can be changed while still passing the specified contours.