Solving occlusion indeterminacy in chromatically homogeneous patterns

Area of Science:

• Visual Perception
• Computational Neuroscience
• Psychophysics

Background:

• Overlapping figures create depth stratification without T- or X-junctions, posing a challenge for determining which surface is in front.
• Previous research has not fully explained the rules governing depth stratification in such ambiguous visual stimuli.

Purpose of the Study:

• To investigate the principle that the visual system minimizes the formation of interpolated modal contours in depth perception.
• To determine if geometrical properties of overlapping surfaces, specifically boundary length, influence perceived depth.

Main Methods:

• Conducted formal experiments and demonstrations involving stereopsis, motion, transparency, motion in depth, and reversible figures.
• Tested the hypothesis that larger surfaces are perceived as closer to minimize occluding boundary length.
• Developed a computational strategy to extract subjective contours from homogeneous patterns.

Main Results:

• The principle of minimizing interpolated modal contours was validated as a determinant of depth stratification.
• Larger surfaces are preferentially seen as closer, correlating with shorter occluding boundaries, independent of relative size cues.
• A computational model successfully replicated human perception of occluding contours in homogeneous patterns.

Conclusions:

• The visual system employs a strategy to minimize contour formation, influencing depth perception in overlapping figures.
• Geometrical constraints, specifically boundary length minimization, play a crucial role in resolving depth ambiguity.
• This principle offers a unified explanation for depth stratification in various visual conditions and can be computationally modeled.