20.500.12556/DKUM-18413 Postopek brezizgubnega stiskanja razčlenjenih vokselskih podatkov Lossless compression of segmented voxel data Doktorska disertacija obravnava področje stiskanja vokselskih podatkov. V uvodu naloge opredelimo problem stiskanja vokselskih podatkov, opišemo cilje naloge in podamo hipoteze, ki jih želimo preveriti. Uvodu sledi opis in definicija vokselskih podatkov, opis naprav, s katerimi podatke pridobivamo ter opis Hounsfieldove lestvice, s katero si lahko pomagamo pri razčlenitvi vokselskih podatkov. V nadaljevanju opišemo področje vokselske grafike Nalogo nadaljujemo s pregledom metod, ki se ukvarjajo s stiskanjem vokselskih podatkov. Opišemo dve metodi, ki sta bili objavljeni v zadnjem času, in podamo nekoliko podrobnejši opis metode QT-B. V jedru doktorske disertacije opišemo podporne tehnike, ki smo jih uporabili pri razvoju naše metode LoCoVox. Podrobneje opišemo standard JPEG-LS in standard JBIG. Sledi podrobnejši opis razvite metode LoCoVox ter njene spletne implementacije VoxelServer, VoxelClient in VoxelDecompressor. V predzadnjem poglavju lastnosti metode LoCoVox analiziramo z eksperimenti. Z metodo smo stisnili nabore vokselskih podatkov in rezultate primerjali s splošnonamenskima programoma ZIP in RAR ter z domenskospecifično metodo QT-B. Metoda je občutno boljša od metode QT-B in ZIP ter primerljiva z metodo RAR, v kolikor želimo prenesti celoten nabor podatkov. Če je dovolj, da prenesemo samo del podatkov (na primer, tkivo, tekočine, kosti), je metoda LoCoVox občutno uspešnejša. The dissertation covers the area of voxel data compression. In the introduction we define the problem of compressing voxel data, describe the goals of the work and state the hypotheses we wish to confirm. The introduction is followed by the description and definition of voxel data, description of the devices for acquiring the data and description of Hounsfield’s scale for analyzing voxel data. In continuation we cover the area of voxel graphics and methods for visualization of voxel data. Next, an overview of methods for voxel data compression is given. We describe two methods that have been recently published and provide a more detailed description of the QT-B method. In the core of the dissertation we describe the support techniques that we applied at the development of our method LoCoVox. We describe in details the JPEG-LS and JBIG standards together with LoCoVox and its web implementation VoxelServer, VoxelClient, and VoxelDecompressor. The next chapter presents experimental verification of LoCoVox. We applied the method to compress voxel datasets and compared the results to the general-purpose compression algorithms ZIP and RAR and the domain-specific method QT-B. The main advantage of our method is that it allows transferring designated areas of data. The method is significantly improved compared to QT-B and ZIP and is comparable to the RAR method in case we want to transfer the complete dataset. If it is sufficient to transfer only a part of data (for instance tissue, liquids, bones), the LoCoVox is considerably more efficient. algoritmi stiskanje podatkov brezizgubno stiskanje vokselski podatki segmentacija algorithms data compression lossless data compression voxel data segmentation true true false [D. Špelič] Slovenski jezik Angleški jezik Doktorska disertacija 2011-05-10 13:03:51 2012-01-31 14:25:02 2022-04-13 10:42:50 2023-12-23 03:37:31 2011 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.925(043.3) 14934806 URN:SI:UM:DK:IOCBAWMJ DR_Spelic_Denis_2011.pdf DR_Spelic_Denis_2011.pdf 1 1B42BAD0A27EA889D97793DFA2F88121 cf10bae1c6129b6972704a5aa18ec33d844cd432b796ebbeb2ec9aba6bdbcdae 1fd6301f-7c05-11eb-bb7a-00155d0001ca 20.500.12556/dkum/b3f04a2e-e08c-442c-bb66-a6b935307209 https://dk.um.si/Dokument.php?lang=slv&id=21918 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0