Interpolacijske metode za upodabljanje oblakov točk

KORDEŽ, JAKA (author), Marolt, Matija (mentor), Bohak, Ciril (comentor)

S tem, ko naše naprave postajajo pametnejše, naraščajo potrebe po zajemanju okolice v digitalno obliko. Poleg načinov samodejne analize teh podatkov je pomembna tudi njihova vizualizacija za uporabnika. Učinkovit algoritem, ki v realnem času izriše prepričljivo rekonstrukcijo skeniranega modela, je pri tem ključnega pomena. V tem delu je opisanih pet pristopov za upodabljanje oblakov točk. Nekateri so enostavni in le uporabljajo funkcionalnosti grafičnih kartic, drugi pa poskušajo izboljšati upodobitev s pomočjo interpoliranja barv. Iz pristopov izhaja sedem različnih implementacij v okolju Rendercore. Opisan je tudi predlog novega algoritma, ki poskuša odpraviti nekatere pomankljivosti ostalih. Vsem osmim algoritmom je izmerjen čas upodabljanja, rezultat pa je primerjan z mrežno rekonstrukcijo modela. Pri tem predlagani algoritem doseže najboljši rezultat v kvalitativnem smislu.

• Language: Slovenian
• Keywords: računalništvo in informatika, računalniška grafika, oblaki točk
• Work type: Master's thesis/paper
• Typology: 2.09 - Master's Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2020
• PID: 20.500.12556/RUL-122197
• COBISS.SI-ID: 41019395
• Publication date in RUL: 26.11.2020

KORDEŽ, JAKA, 2020, Interpolacijske metode za upodabljanje oblakov točk [online]. Master’s thesis. [Accessed 31 March 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=122197

Secondary language

• Language: English
• Title: Interpolation methods for point cloud rendering

Abstract

As our electronic devices are becoming more sophisticated, the needs to capture the environment in a digital form are growing. In addition to the ways of automatically analysing the data, the visualization of the respective data is also important for the user. Therefore, an efficient algorithm that plots a convincing reconstruction of the scanned model in real time is essential. This thesis describes five approaches to rendering point clouds. Some of the approaches are simple and only use the functionality of graphics cards, while others try to improve the rendering by interpolating colours. Seven different implementations in the Rendercore framework are derived from the approaches. A proposal for a new algorithm, which tries to eliminate some of the shortcomings of the others, is also described. Rendering time is measured for all eight algorithms, and the result is compared with a mesh reconstruction of the model. In qualitative terms, the proposed algorithm achieves the best results.

• Keywords: computer science, computer graphics, point clouds