多分辨率体绘制数据动态加载方法的研究

发布时间：2018-06-06 04:44

本文选题：多分辨率 + 体绘制　；参考：《吉林大学》2017年硕士论文

【摘要】：随着可视化技术的持续发展和进步,在越来越多的领域都将这项技术用于实践当中,尤其在地质探测领域。大规模地学数据的可视化提供了地球内部的真实情况,通过三维地震数据可视化技术,我们可以直观观察到体数据的内部结构和属性,并可以通过渲染形成的清晰图形图像对地震数据进行分析和解释,而不是单纯依靠数值表格。随着技术的不断成熟进步,地质勘探按领域中的实验数据程快速增长达到几十GB甚至TB级别,在进行数据可视化时地震数据不能完全加载入内存当中进行处理。传统的将体数据完全加载入内存当中,渲染模块只从内存中获取数据的渲染模式在海量三维地震数据渲染过程中效率很低。因此作为海量数据体绘制过程中的关键技术,深入研究基于层次细节分块多分辨率体绘制技术很有必要,同时对大规模三维体数据的存储以及加载过程的管理也是需要解决的问题。多分辨率体绘制技术的算法首先对数据体进行划分,并找到其中的同质区域。算法对这些区域进行采样,通过对采样的分析,算法获得体绘制过程中不同分辨率的标量数据信息,可以根据通过采样分析得到的数据信息确定细节分块水平,这样可以有效的降低用于体绘制的数据量。本文详细研究和分析了三维数据体绘制的方法,包括直接体绘制方法、并行体绘制方法以及多分辨率体绘制方法,并对其中多分辨率体绘制算法中层次划分的八叉树表示和统一层次划分的多分辨率表示进行详细分析和对比。同时基于大数据背景,分析研究了大规模数据的数据存储、内存管理以及数据加载等大数据处理流程,将所有研究内容用于对三维地学数据可视化工具包Open Probe中的部分模块进行优化。在基于Open Probe地学项目的背景下,文章针对Open Probe中数据加载效率以及分辨率变化过程中图像显示不流畅、没有动态性的问题提出解决和优化方案。文章研究分析了LDM文件格式,并和传统地震数据文件格式SEG-Y进行比较分析。通过对大规模体数据八叉树结构存储方式以及大规模数据快速加载机制的研究,设计实现多线程加载,多优先级缓存替换策略的加载机制,在内存的替换策略中设立标志位机制,保证数据块替换的合理性。同时,通过对基于八叉树层次划分结构的多分辨率体会绘制方法的研究,在用户交互过程中,分辨率发生变化时,采用双线程双链表的方法结合标志位替换机制,实现了在分辨率变化时体数据渲染与显示的动态性和流畅性,对Open Probe原有的渲染显示过程进行一定程度的优化。
[Abstract]:With the continuous development and progress of visualization technology, this technology has been applied in more and more fields, especially in the field of geological exploration. The visualization of large-scale geoscientific data provides the real situation of the earth's interior, and through the three-dimensional seismic data visualization technology, we can visually observe the internal structure and properties of the volume data. The seismic data can be analyzed and interpreted by rendering clear graphics instead of relying solely on numerical tables. With the development of technology, the experimental data process of geological exploration in the field of rapid growth to tens of GB or even TB level, seismic data can not be completely loaded into memory for processing in data visualization. The traditional rendering mode which only gets data from memory is inefficient in the process of massive 3D seismic data rendering. Therefore, as a key technology in the process of massive data volume rendering, it is necessary to deeply study the multi-resolution volume rendering technology based on hierarchical details. At the same time, the storage of large-scale 3D volume data and the management of loading process are also problems to be solved. The algorithm of multi-resolution volume rendering firstly partitioned the data volume and found the homogenous region. These regions are sampled by the algorithm. By analyzing the sampling, the algorithm obtains scalar data information of different resolution in the process of volume rendering, and can determine the level of detail block according to the data information obtained by sampling and analysis. This can effectively reduce the volume of data used for volume rendering. In this paper, the methods of 3D volume rendering, including direct volume rendering, parallel volume rendering and multi-resolution volume rendering, are studied and analyzed in detail. The octree representation of multi-resolution volume rendering algorithm and the multi-resolution representation of unified hierarchical partitioning are analyzed and compared in detail. At the same time, based on the background of big data, the big data processing flow of large-scale data storage, memory management and data loading is analyzed and studied. All the research contents are used to optimize some modules in 3D Geoscience Visualization Toolkit (Open Probe). Based on the background of Open Probe geoscience project, this paper proposes a solution and optimization scheme for the problems of data loading efficiency and resolution change in Open Probe, in which the image display is not smooth and dynamic. In this paper, the LDM file format is analyzed and compared with the traditional seismic data file format SEG-Y. Based on the research of large volume data octree structure storage and fast loading mechanism of large scale data, the loading mechanism of multi-thread loading and multi-priority cache replacement strategy is designed and implemented. In order to ensure the rationality of data block replacement, a flag bit mechanism is set up in the memory replacement strategy. At the same time, through the research of multi-resolution experience rendering method based on octree hierarchical partition structure, in the process of user interaction, when the resolution changes, double thread double linked list method combined with flag bit replacement mechanism is adopted. The dynamic and smooth rendering and display of volume data is realized when the resolution changes, and the original rendering and display process of Open Probe is optimized to a certain extent.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631.4;TP391.41

【相似文献】