Thesis Details
Zobrazení rozsáhlých volumetrických dat na CPU
This thesis deals with the problem of displaying volumetric data that exceeds the operating memory capacity of the machine. The work describes the design of a visualization pipeline, which consists of a data structure for large volumetric data and an algorithm that visualizes such data. The proposed hierarchical data structure accelerates sampling and allows the reduction of the total amount of data that needs to be loaded into physical memory during visualization. Visualization of processed data is achieved by the ray casting method with existing optimization techniques, such as empty space skipping and early ray termination. The data structure allows up to 12x faster sampling compared to the sampling of raw large volumetric data serialized by rows. Up to 150x faster visualization of large volumetric data in near-lossless mode has been achieved compared to the fully lossless mode by utilizing the data hierarchy. The display scheme is implemented in the form of a library in C++20 language. The implementation uses acceleration by vectorization and allows easy parallelization by the user. The library provides tools for processing and visualization of large volumetric data on the CPU.
Volumetric data, large data, block decomposition, ray casting, vectorization, data hierarchy, octree.
Bařina David, Ing., Ph.D. (DCGM FIT BUT), člen
Češka Milan, doc. RNDr., Ph.D. (DITS FIT BUT), člen
Grézl František, Ing., Ph.D. (DCGM FIT BUT), člen
Milet Tomáš, Ing., Ph.D. (DCGM FIT BUT), člen
Zemčík Pavel, prof. Dr. Ing. (DCGM FIT BUT), člen
@mastersthesis{FITMT24487,
author = "Drahom\'{i}r Dlabaja",
type = "Master's thesis",
title = "Zobrazen\'{i} rozs\'{a}hl\'{y}ch volumetrick\'{y}ch dat na CPU",
school = "Brno University of Technology, Faculty of Information Technology",
year = 2022,
location = "Brno, CZ",
language = "czech",
url = "https://www.fit.vut.cz/study/thesis/24487/"
}