event

Ph.D. Defense of Dissertation: Jiajian Chen

Primary tabs

Title: Non-photorealistic Rendering with Coherence for Augmented Reality
 
Jiajian Chen
School of Interactive Computing
College of Computing
Georgia Institute of Technology
 
Date: Friday, June 29th, 2012
Time: 8:00 AM - 10:30 AM
Location: Technology Square Research Building (TSRB) Room 223
 
Committee

  • Dr. Greg Turk (Advisor, School of Interactive Computing, Georgia Institute of Technology)
  • Dr. Blair MacIntyre (Advisor, School of Interactive Computing, Georgia Institute of Technology)
  • Dr. Frank Dellaert (School of Interactive Computing, Georgia Institute of Technology)
  • Dr. Irfan Essa (School of Interactive Computing, Georgia Institute of Technology)
  • Dr. Eugene Zhang (Electrical Engineering and Computer Science, Oregon State University)

 
Abstract
A seamless blending of the real and virtual worlds is key to increased immersion and improved user experiences for augmented reality (AR). Photorealistic and non-photorealistic rendering (NPR) are two ways to achieve this goal. Non-photorealistic rendering creates an abstract version of both the real and virtual world by stylization to make them indistinguishable. This could be particularly useful in some applications, e.g., AR/VR aided machine repair, or for certain AR games with artistic stylization. Temporal coherence is a key problem for all NPR algorithms. Rendered results are temporal coherent when each frame smoothly and seamlessly transitions to the next one without visual flickering and artifacts that distract the eye from smoothness. In this thesis we present three NPR algorithms with coherence for AR, two that are image space algorithms and one that is a mode space algorithm. We utilize the information from both the real content and virtual content to maintain coherence in our algorithms.
 

  1. We present a watercolor inspired NPR algorithm for AR. This algorithm produces a watercolor-like visual effect for the AR videos. Coherence is achieved by computing the center average of edge pixels in each Voronoi cell and re-tilling Voronoi cells based on these edge centers. Our contribution is using Voronoi diagrams to produce the water-color like rendering style with coherence at interactive frame rates.
  2. We present a painterly rendering algorithm in image space for AR. This algorithm renders AR graphics content with a paint and brush style. Coherence is achieved by warping brush stroke anchors from frame to frame and placing brush strokes at these anchor positions. Each brush stroke is also re-shaped to provide better coherence in the final rendering. Our contribution is a new method of creating tensor fields and warping brush anchors between frames.
  3. We present an NPR framework with support of coherence in model space for AR. Our systems targets painterly rendering styles of NPR. The contribution of our algorithm has two parts. The first part is the anchor sampling algorithm. This algorithm maintains proper density of brush anchors on the screen, which is a desired feature for many NPR algorithms. It is also particularly suitable for AR, as we do not know the camera motion in advance. The second contribution is our method of averaging brush properties, including their skeletons and colors, to achieve better coherence in the final rendering. Compared with existing methods, our method allows us to smoothly blend curly brush strokes with a cubic B-spline representation. We apply these methods to both static and animated models to create a painterly rendering style for AR. Compared with existing image space algorithms our method renders AR with NPR effects with a high degree of coherence.

Status

  • Workflow Status:Published
  • Created By:Jupiter
  • Created:06/28/2012
  • Modified By:Fletcher Moore
  • Modified:10/07/2016

Categories

  • No categories were selected.

Keywords

  • No keywords were submitted.