Please use this identifier to cite or link to this item: https://www.um.edu.mt/library/oar/handle/123456789/91577
Title: Temporal filtering for minimizing artifacts in progressive rendering of animations
Authors: Camenzuli, Maria (2014)
Keywords: Hypertext systems
Image processing
Optical data processing
Issue Date: 2014
Citation: Camenzuli, M. (2014). Temporal filtering for minimizing artifacts in progressive rendering of animations (Bachelor's dissertation).
Abstract: Progressive rendering, a technique that mitigates the high latency in frame generation by providing visual feedback to the user as early as possible, then gradually refining its quality, introduces noisy artifacts into the animation of dynamic scenes, degrading the quality of the animation. This technique is one of the many creative solutions that developers of real-time rendering applications have come up with in an attempt to maintain the balance between interactivity and high visual quality. Another such solution is the exploitation of temporal coherence, were we take advantage of the fact that consecutive frames in interactive rendering exhibit very little difference in point shading. This report presents our investigation into the use of temporal filtering to improve image quality in progressive rendering, independent of the occurrence of scene changes. We start by analyzing temporal coherence in scenes that are dynamic in terms of camera movement, which we accomplish through the implementation of temporal frame reconstruction. We then proceed to implement a temporal filter that combines temporal data with the information in the target frame using a weighted average. Our evaluation results show that, when enough temporal data is available, temporal reconstruction is capable of creating high quality frames with very little perceivable differences when compared to the corresponding original frames, while lack of temporal data creates very distracting gaps in the reconstructions which would have to be traditionally rendered for the frame to be complete. The results also demonstrate the effectiveness of our temporal filter when applied to noisy frames, achieving significant improvements in visual quality. However, applied to frames that are not so noisy to begin with, a degree of blurring is introduced into the image, degrading quality. Overall, we see potential in our proposed solution, and therefore propose further work to build upon what we have done.
Description: B.SC.(HONS)COMP.SCI.
URI: https://www.um.edu.mt/library/oar/handle/123456789/91577
Appears in Collections:Dissertations - FacICT - 2014
Dissertations - FacICTCS - 2010-2015

Files in This Item:
File Description SizeFormat 
BSC(HONS)ICT_ Camenzuli_Maria_2014.PDF
  Restricted Access
10.83 MBAdobe PDFView/Open Request a copy


Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.