Estimation of 3D shape from shading and binocular disparity

Dövencioğlu, Dicle Nahide (2013). Estimation of 3D shape from shading and binocular disparity. University of Birmingham. Ph.D.

Preview

Dovencioglu13PhD.pdf
PDF - Accepted Version
Download (2MB)

Abstract

How does the visual system make use of various sources of information to the three-dimensional (3D) geometry of the world? To infer distances in a 3D scene, the brain uses multiple cues such as binocular disparity, which provides metric estimates of depth; or shading, which is inherently ambiguous and requires additional interpretation. In this thesis, I use psychophysical and functional magnetic resonance imaging (fMRI) techniques to address the following questions: (i) how does the visual system resolve ambiguities in a luminance signal, especially, separating shading cues to shape from luminance variations caused by the changes in the surface material, (ii) when both shading and binocular disparity are available, how do these cues interact to produce a coherent 3D shape estimate, (iii) what is the neural substrate to this cue integration?

First, in Chapter 3, I examine how first- and second-order luminance signals in a luminance pattern are perceived, and ask if observers can benefit from the phase relationship of these signals as a cue to shape. Next, in Chapter 4, I ask whether decomposing shading and reflectance cues to infer shape can be done in very short presentation times. In Chapter 5, I present evidence that the involvement of V3B/KO in 3D shape processing can be extended to disparity and shading signals. Moreover, I find a distinct relation between neural activity in this cortical area and perceptual judgments of individual observers. Finally, in Chapter 6, I carry on investigating cue integration to gain further insight onto the individual variations.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):