Visual accessibility is "the effectiveness with which vision can be used to travel
safely through the space and to pursue the intended activities in the space (Legge, Yu,
Kallie, Bochsler, & Gage, 2010)." A major issue for visual accessibility involves the
visibility of objects under visually challenging conditions, especially those faced by
people with impaired vision. This research investigates visibility of objects under
conditions of artificially induced blur, and concludes with simple design principles for
improving object visibility when vision is compromised by reductions in acuity and
contrast sensitivity. Additionally, this research bridges a gap between object visibility
studies in physical environments, and studies of visibility using image processing and
screen displays of the same objects and scenes. Psychophysical measurements of
visibility of objects shown on the desktop display accounted for 58% of the variability in
visibility measurements of the same scenes and objects in the physical environment. The
technical aspects of virtual modeling—from stimulus conception to photometrically and
geometrically accurate displaying of stimulus images—are described.
Chapter 1 of this dissertation provides an overview of this research.
In Chapter 2, experiments are described in which object visibility was tested in a
physical environment. The objects were boxes and cylinders. The subjects were normally
sighted young adults who wore blur goggles that simulate the effects of visually impaired
acuity and contrast sensitivity loss. Interactions between illumination source (i.e.,
luminaire location), surface reflectance, and object shape led to the conclusion that curved objects were more visible than flat objects with blurry vision. However we wished
to confirm this conclusion for a wider range of viewing conditions.
Chapter 3 extended the work of the physically based study to a computer desktop
displayed environment. Several models for digitally simulating physical blur goggles
were examined. A high dynamic range (HDR) image of the point spread function (PSF)
of the physical blur goggles, convolved with HDR images of the stimuli, plus a slight
contrast attenuation, provided the best agreement between visibility performance of
subjects viewing the objects on a computer display and the performance of the subjects
who viewed the corresponding real objects (Chapter 2). A comparison between
photographs and renderings showed no difference between the two image types. The main effects and interaction findings of the physical experiment (from Chapter 2) were
largely reproduced with our desktop model. The desktop model was used to study novel
conditions (i.e., new object locations and orientations), were it was found that under
diffuse overhead illumination near the center of the room, the curved objects were
actually less visible than flat objects. In all other cases that were tested, curved objects
still provided enhanced visibility over flat objects.
Chapter 4 contains a technical report on the methods used to create photometrically
accurate HDR images based on physically based 3D objects, materials, and scenes. It
illustrates the 3D modeling, rendering, photography, and photometry methods for
reproducing photometrically and geometrically accurate scenes on a calibrated computer
display. In addition to modeling and calibration techniques, Chapter 4 illustrates methods
for spatial filtering (i.e. visual blur), and non-spatial image processing, including color balancing and contrast reduction, required for accurate reproduction of the physically
based experiment. Example code for all major steps, including spatial and non-spatial
filtering, is provided.
Collectively, this series of experiments demonstrated a simple heuristic for designing
visually accessible spaces: namely that adding curvature to objects can enhance object
visibility under reduced acuity and reduced contrast sensitivity situations. This
enhancement, however, is lost under diffuse illumination, where other methods for
increasing contrast within or around the object would be more reliable.
Moreover, this research demonstrates a complete workflow from conception of a
physical environment to complete virtual modeling of the physical space, as well as
digitization of models for visual impairment. This research provides a foundation for
further study of visibility in more complex and realistic environments (including motion,
as demonstrated in this dissertation's supplemental video archive), linking virtual desktop models to physical environments.
University of Minnesota Ph.D. dissertation. December 2012. Major: Psychology. Advisors: Gordon E. Legge, Paul R. Schrater. 1 computer file (PDF), xvi, 138 pages.
Kallie, Christopher Scott.
Evaluating object visibility with simulated visual impairment using real and rendered scenes.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.