中文摘要：

图形硬件的发展为实时体数据可视化提供了硬件保证,然而随着扫描技术的发展,大数据可视化仍然面临显存不足问题,因此研究保持数据特征的压缩表达方法就非常重要。应用张量近似思想建立了体数据的多尺度表达与可视化方法,一方面多尺度张量近似实现了数据压缩,解决了大数据的绘制问题;另一方面,张量近似的自适应压缩基保持了体数据的尺度特征。实验结果表明,该方法是有效的。

英文摘要：

Real time volume visualization is supported by modern graphics hardware currently.However,with the development of scanning technology,the performance in GPU memory is still a bottleneck for large volume data visualization.Thus it is very important to find a compression method which can preserve volume feature.It is developed based on tensor approximation.The approach can be used for data compression based on multi-scale tensor approximation.At the same time,volume feature is preserved since compression base is extracted from data.The obtained results show that the approach is valid.