Watson, Richard Stewart; (2002) Complex motion processing. Doctoral thesis (Ph.D), UCL (University College London).

Text Complex_motion_processing.pdf Download (14MB)

Abstract

When an observer moves, or objects in his field of view move, complex patterns comprising a heterogeneous range of speeds and directions are projected onto the retinae. Several classes of these optic flow patterns have been identified, including translation, rotation, radiation and shear. This thesis investigates the properties of motion-sensitive detectors in human vision using established psychophysical techniques. The range of directions to which rotation and radial motion detectors are sensitive were measured. These 'bandwidths' were found to be narrower than estimated by previous investigators, a result that is incompatible with the notion that complex patterns are coded by a basis set of elementary components. The size and spectral content selectivity of expansion detectors was measured. Surprisingly the visual system prefers pattern elements not to change size as they approach, contrary to real life. Similarly, optimal performance was found with constant spatial frequency, except at high speeds. We conclude that the spatial frequency tuning of low-level motion detectors places a fundamental limit on the detection of radial patterns. A cue conflict paradigm was used to explore the exaggerated speed experienced with radial motion. Additional depth cues were ineffective in compromising the illusion, suggesting that it may not be due to a motion-in-depth interpretation, or that cue combination in the visual system weights relative motion more highly than other depth cues. The smallest difference in speed that can be discriminated in rotation, radial motion and translation patterns was measured using two techniques. We conclude that thresholds are in keeping with existing literature, though results were inconsistent between techniques. The meaning of some forms of motion is ambiguous without additional information. We tested whether sensitivity to shear motion is enhanced when environmentally relevant depth information is added through binocular disparities. Contrary to previous findings depth information plays no part in shear detection.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as XMLJSONCSV

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item