All ETDs from UAB

Advisory Committee Chair

Allan Dobbins

Advisory Committee Members

Claudio Busettini

Timothy Gawne

Thomas Norton

Andew Pollard

Donald Twieg

Document Type

Dissertation

Date of Award

2006

Degree Name by School

Doctor of Philosophy (PhD) School of Engineering

Abstract

A fundamental problem in binocular vision is to understand the rules that govern matching of features in the two eyes. We adopted Panum’s limiting case (PLC) into random-dot stereograms (RDS) to determine whether our visual system permits non-unique binocular matching for stereopsis. Using PLC RDS, we found that i.) the PLC-defined depth was compatible with the disparity-defined depth, which suggests that our visual system detects the relative disparity in PLC; ii.) the upper depth limits discriminating the PLC-defined surfaces and disparity-defined surfaces were almost identical, which suggests that there is a common mechanism – correlation-based stereopsis – responsible for the depth phenomenon. Furthermore, when one of the two matches in PLC was anti-correlated (of opposite contrast), the PLC RDS induced reversed depth, which is consistent with the neuronal and visuomotor responses. The reversed depth phenomenon is an indication that ordinary disparity detection produces the anti-correlated signal in PLC that is perceived as depth, an exception for anti-correlated stimuli. To explain why we experience depth reversal in the anti-correlated PLC RDS but not in the conventional anti-correlated stimulus, we simulated the response of binocular disparity-sensitive neurons. We found that the disparity estimates in the correlated RDS were coherent across scale, whereas the disparity estimates were dispersed in anti- ii correlated RDS. In contrast, the dispersion of the disparity estimates was limited to a range of spatial frequencies in the anti-correlated PLC RDS. Evidently, confined estimates were coherent enough to yield the sign of depth, but not coherent enough to support surface integration. In addition, we found that limiting the spatial frequency channels available for disparity detection reduced the depth discrimination accuracy. Based on these observations, we propose a second stage mechanism that combines the elementary signals of the disparity detectors to produce the disparity map that resolves transparency in depth surfaces. The second stage mechanism incorporated the facts that 1) the removal of the monocular components from the elementary signal sharpens the disparity tuning; 2) inter-scale summation improves the signal-to-noise ratio and broadens the working disparity range. This model network further predicted the degree of decoherence in the anti-correlated RDS and the reversed depth phenomenon.

Included in

Engineering Commons

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.