Research and Collaborators

PDF: Research Statement & Statement of Scholarly Impact
Current Research | Graduate Students | Undergraduate Students | Funding | Collaborators

    Current Research

      Image Creation and Manipulation.

        The creation and modification of imagery must be done with care, otherwise the content and context of the information may be misconstrued or lost. In addition to data-reduction and sampling errors introduced by some image manipulations, modification of imagery may not preserve the semantic content or important large scale features of an image.

        For example, consider converting an image from color to grayscale. Default grayscale conversion algorithms remove the visible boundaries between isoluminant colors. In my research, I provide a solution to this problem that incorporates the perceived boundaries in the color image into the grayscale image [Gooch et al. 2005]. The results of this method show that careful image manipulations that preserve perceptual differences create improved grayscale images and are useful when it is not possible to reproduce a full color image, such as on a grayscale printer. In addition to being useful for improving grayscale printer drivers, the techniques used in our Color2Gray method may be useful for detecting camouflage or possibly even creating better camouflage. Based upon the momentum provided in this work, several other researchers are examining speed-ups to the algorithm, as well as exploring the technique for color-deficient viewers. I am currently exploring approaches for recoloring badly lit photographs containing people on the simple premise that human skin has a limited range of valid colors [Long and Gooch 2009].

       

      Multispectral Computational Photography

        I am creating a color science lab that is not restricted to purely visual wavelengths. Such a lab will include a closed loop setup of input and output devices and a perceptual experimentation workstation.
        I have a broad interest in distilling and presenting a lot of information in a single image. For example, think about how a technical illustration distills the key elements of a CAD part to a simplified drawing. I want to go from captured spectra to images that illustrate the perceptually salient information in the image, not just spectra to colored pixels. In this process, we can also consider folding in non-visible spectra as well. [Long et al. 2010] [Olsen et al. 2010]

        The idea:

        Current State of our art:


       

      Computational Aesthetics and Non-Photorealistic Rendering

        The advent of photography and computer graphics has not replaced artists. Imagery generated by artists provides information about objects that may not be readily apparent in photographs or real life. The same goal should apply to computer-generated images. This is the driving force behind non-photorealistic rendering. The term non-photorealistic rendering (NPR) is applied to imagery that looks as though it was made by an artist, such as pen-and-ink or watercolor. Many computer graphics researchers are exploring NPR techniques as an alternative to photorealistic rendering. More importantly, non-photorealistic rendering is now being acknowledged for its ability to communicate the shape and structure of complex models. Techniques which have long been used by artists can emphasize specific features, expose subtle shape attributes, omit extraneous information, and convey material properties. These artistic techniques are the result of an evolutionary process, refined over centuries. Therefore, imitating some of these artistic methods and exploring the perceptual psychology behind the techniques of artists are good first steps in going beyond photorealistic rendering.

        In my research [Gooch et al. 1998] [Sloan et al. 2001] [Gooch et al. 1999] [AlMeraj et al. 2009] [Mehra et al. 2009], I have demonstrated how non-photorealistic rendering methods can be used to convey a more accurate or appropriate representation of the shape and material properties of objects than traditional computer graphics methods, specifically in computer-generated technical illustrations. There are four local shading models in computer graphics, each named for the investigator who discovered it; Phong, Gouraud, Blinn and Gooch [Gooch et al. 1998]. The Gooch shading model, first presented in 1998 has been successfully used in technical and medical illustration, as well as making an impact on the visuals for Walt Disney Feature Animation movies like ”Treasure Planet”. Additionally, I co-authored the first book [Gooch and Gooch 2001] published on the topic of ”Non-Photorealistic Rendering” and am working on a follow-up book that explores the new trends in non-photorealistic rendering and illustrative visualization.
         

        Casual Educational Games.


          I am exploring creating casual, entertaining, educational games that provide additional learning opportunities for a wide audience of players. Casual gaming supports a lack of commitment of time or often requires a reduced set of skills required by hard core games. In collaboration with undergraduate and graduate students, we have designed a few educational games that are considered entertaining, such that the content itself is engaging and appeals to the players. Such games are created for simple mouse interfaces as well as single and multi-touch surfaces. The goal is to create games or interactions that, within a minute or two of playing, will communicate at least as much information as would be conveyed in the typical information placards found near zoo or museum exhibits. Example games include "Catch that Frog", an interactive projector-based game about frogs and camouflage, and "Catalyst", a simulation of the cat visual system that is based on the research of neuroscientists [Long et al. 2010] My current research is focused on exploring other vision systems (bee, pit-viper) in search of image and interaction mechanisms for displaying information beyond the visible spectrum in a single image.


         

        Human Perception in Immersive Environments.

          Although, non-photorealistic rendering (NPR) methods allow us to emphasize or omit detail in order to communicate information more effectively, an important issue to consider is how can we evaluate and validate the communication adequacy of these techniques? I presented the first paper in the field of NPR [Gooch and Willemsen 2002] that provides a methodology for quantitatively evaluating spatial perception in a functional, NPR, stereo, immersive environment. The methods described in this paper provides NPR researchers with a methodology for conducting user studies to perceptually validate new rendering approaches within immersive environments. I have also explored various approaches for generating photorealistic stimuli and examined the effect they have on the same kind of action-based tasks in immersive environments [Creem et al. 2005] [Creem et al. 2004] [Hu et al. 2002] [Thompson et al. 2004] [Willemsen et al. 2002]. Such investigations have a broad range of implications for the utility and practicability of visually immersive interface for applications such as simulation, education and training.


         

      Graduate Students

        Jeremy Long, PhD student [Old USask Web page]
        Anthony Estey, MS student
        Ji Li, MSc Student
        Ryan Wallace, MSc Student
        Shelley Gao, MSc Student

       

      Undergraduate Students


       

      Research Collaborators

        University of Victoria

        The University of Victoria Island Graphics Group, newly established in Fall 2006, consists of myself as well as Dr. Brian Wyvill (CRC, Computer Science), Dr. Bruce Gooch, and Dr. Melanie Tory and their students.

        Paul Lalonde (Intel, Adjunct Prof. at UVic), and Jean-Luc Duprat (Intel)

        At the University of Victoria, I am actively collaborating with Visual Arts Professor in Sculpture, Prof. Lucy Pullen on a project that seeks to improve the ability to created realizable 3D structures in the built world from 3D computer generated models, typically only visualizable in 2D.

        Nationally

        Collaborations are underway between myself and several researchers in Canada including Dr. Wolfgang Heindrich (UBC), Dr. Alla Sheffer (UBC), and Dr. Sheelagh Carpendale (Calgary), Dr. Mario Costa-Sousa (Calgary).

        Internationally

        Dr. Gooch maintains strong ties and active collaborations with academic institutions and researchers around the world including perceptual psychologists such as Dr. Herb Pick (University of Minnesota, Twin Cities), Dr. John Rieser (Vanderbilt University), Dr. Sarah Creem-Regehr (University of Utah), with regard to human task performance in virtual environments. Additionally she has works with psychologists Dr. Marcia Grabewecky (Northwestern University) and Dr. Peter Vishton (William and Mary) on carrying out perception experiments on computer-generated imagery.

        Other collaborators include: Dr. Pete Willemsen (University of Minnesota Duluth), Dr. William B. Thompson (University of Utah), Dr. Neil Dogdson (Cambridge, UK) and Dr. Ivan Viola (Bergen, Norway).

        She has also collaborates with artists such as Prof. Maureen O'Hare-Ur (University of Utah), creating cross-disciplinary graduate research seminars and often consults with artist and painter Susan Ashurst (Utah).


       

      Funding

        Grants:

        • Networks of Centres of Excellence of Canada, GRAND: Graphics, Animation, and New Media , Cash , $11,000/year
        • & Advanced Graphics Hardware Research & April 2009 -- April 2010, $35,000
        • , Semantic Icons & Feb. 2009, $8500
        • Colour Science for Computational Photography, May. 2008 -- May 2013, $178,000
        • NSERC USRA & Art and CS Collabroation (Shelley Gao), May. 2009 -- Aug. 2009, $4500
        • Advanced Graphics Hardware Research, April 2008 -- April 2009, $35,000
        • Save the Frogs: Entertaining Educational Games for Zoo Exhibits (Donna Kaminskyj, Elyse Regan, Kyliah Clarkson),Blog
        • NSERC USRA & Games for CS (Donna Kaminskyj, Elyse Regan), May. 2008 -- Aug. 2008, $9000
        • CRA-W CDMP Ugrad Research, "Pointer for Freshman: Teaching Computer Science Concepts through Educational Games". (Donna Kaminskyj, Elyse Regan) Blog, May. 2008 -- Sept. 2009, $8000
        • NSERC Discovery Grant, Preserving Salience By Maintaining Perceptual Differences for Image Creation and Manipulation, 2006 -- 2012, $90,000
        • Google Earth Competition: UVic Model (Brittany Sims, Susan Perkins), May. 2007 -- Aug. 2007, $9000
        • CRA-W CDMP Ugrad Research "Pointer for Freshman: Teaching Computer Science Concepts through Educational Games". (Susan Perkins) & May. 2008 -- Sept. 2009, $5000
        • UVic Professional Development Faculty Award, May. 2007 -- May. 2008, $2,743
        • UVIC IRG Top-Up Award, May. 2007 -- May. 2008, $675
        • Autodesk Research Donation: 9 seats of Maya Unlimited Rendering and Animation, Cleaner, Combustion, and 3D Studio Max & Sept. 2001 -- Present, In-Kind, $160,000
        • Games on Multiple Hardware Platforms& 2005--2006, $80,000

          Note: Canadian funding is unique in two key ways:

          • Grants do not incur overhead.
          • Industry donations receive 2-to-1 or 3-to-1 matching from federal and provincial sources.


         

        Fellowship:

        • Helen and Robert J. Piros Fellowship 2004-2005


       

      Undergraduate Research Projects Ideas

      • Jeapordy-like program that allows me to set up categories, q&a, three group scores, etc to be used as a learning tool. Also need to create data for it for Intro CS courses: CS history, CSC212 topics, etc.
      • Blog Aesthetics
      • Pentimenti image processing followup
      • Survey of Art with respect to mappings
      • Game and Education