Attentional flow networks : a real-time adaptive display technique Public Deposited

http://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/rx913s69w

Descriptions

Attribute NameValues
Creator
Abstract or Summary
  • Graphics resource allocation has often existed on the periphery of graphics research, taking second place to faster or more realistic algorithms for global illumination and modeling. Most research has focused on generating sets of images that best match the effects one might see in photographs. While this approach works well for digital image synthesis techniques, real-time rendering possesses a fundamentally different character. The core problem in a real-time renderer is designing a system to best allocate resources with respect to a viewer in real time while realism is important to this goal, viewers do not perceive individual frames in a frame sequence in the same way that viewers perceive the images produced by unconstrained synthesis techniques. In this thesis, we present a method for performing graphics resource allocation that employs a new structure called the attentional flow network (AFN). The attentional flow network technique integrates a data structure similar to an artificial neural network with a scene graph. This structure defines how the "relevance" of objects with respect to human viewers flows to other objects in subsequent time periods. At the same time, external activations are imposed on this structure using models of the viewer generated by an external system (the center-of-focus predictor). We explore both a basic version of this technique and a sequence of extensions to the basic technique. Further, while some work has been done on center-of-focus prediction systems, the problem still generates substantial difficulties for techniques that require user information to distribute resources, as the attentional flow network technique does. We thus discuss several approaches that might be taken to modeling the user center of focus. We discuss two new approaches: the warp-and-woof technique and the semantic scene-cell decomposition function approach. The first estimates user focus using a function of the mouse-pointer location and the locations of significant objects. The second estimates user focus using a programmer-defined decomposition of space that includes a coarse sampling of likely user centers of focus given position and orientation of the viewpoint. Finally, we evaluate the basic form of the attentional flow network technique. This evaluation takes the form of three studies that examine the cost of using the technique, the benefits gained versus unconstrained rendering, and the benefits gained versus constrained rendering with uniform resource allocation. Our results suggest that the attentional flow network technique has the potential to substantially improve resource allocation at little cost, particularly in systems that are not CPU-limited.
Resource Type
Date Available
Date Copyright
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Academic Affiliation
Non-Academic Affiliation
Subject
Rights Statement
Peer Reviewed
Language
Digitization Specifications
  • File scanned at 300 ppi (Monochrome, 8-bit Grayscale) using ScandAll PRO 1.8.1 on a Fi-6770A in PDF format. CVista PdfCompressor 4.0 was used for pdf compression and textual OCR.
Replaces
Additional Information
  • description.provenance : Approved for entry into archive by Patricia Black(patricia.black@oregonstate.edu) on 2012-08-23T17:30:28Z (GMT) No. of bitstreams: 1 BibleJohn2001.pdf: 6622491 bytes, checksum: b47590cbcaebdd552713336e8c38d933 (MD5)
  • description.provenance : Made available in DSpace on 2012-08-23T17:30:28Z (GMT). No. of bitstreams: 1 BibleJohn2001.pdf: 6622491 bytes, checksum: b47590cbcaebdd552713336e8c38d933 (MD5) Previous issue date: 2000-08-15
  • description.provenance : Submitted by Kaylee Patterson (kdpscanner@gmail.com) on 2012-08-22T18:02:22Z No. of bitstreams: 1 BibleJohn2001.pdf: 6622491 bytes, checksum: b47590cbcaebdd552713336e8c38d933 (MD5)
  • description.provenance : Approved for entry into archive by Patricia Black(patricia.black@oregonstate.edu) on 2012-08-23T17:25:39Z (GMT) No. of bitstreams: 1 BibleJohn2001.pdf: 6622491 bytes, checksum: b47590cbcaebdd552713336e8c38d933 (MD5)

Relationships

Parents:

This work has no parents.

Last modified

Downloadable Content

Download PDF

Items