Real-time switching and visualization of logging attributes based on subspace learning

Abstract The Three-Dimension visualization effect is limited by the performance of equipment and algorithm when dealing with high-dimensional and large-scale geological data. So it is very difficult to graph the data accurately in real-time. In this paper, an accurate and efficient real-time visualization method is studied, which combines the distribution characteristics of geological data in space and the physical law of multi-dimensional logging attributes. Firstly, the spatial data field is evenly divided into sampling intervals of the same size, and the dispersion degree of numerical distribution of geological data in different sampling intervals is counted, and the sampling interval is divided into different sampling density grades according to the degree of dispersion clustering. The logging data are sampled according to the sampling density level in each sampling interval to compress the data scale. Secondly, the Pearson coefficient is used to classify logging attributes to solve visual switching delay caused by multiple dimensions of data. Third, The basis vector and its coefficients are obtained through intra-class subspace learning, based on which the mapping model between attributes is established. When the attribute is switched, only the coefficient value in the mapping model needs to be changed to reduce the amount of data exchange and ensure real-time visualization. The experimental results show that real-time visualization of large-scale geological data can be realized using this method, which supports multi-attribute dynamic switching and has good rendering accuracy.