The information content of texture gradients

Area of Science:

• Visual perception
• Computational geometry
• Psychophysics

Background:

• Texture gradients systematically vary in projected surface texture.
• Image variables like texture density and size are expected to convey surface orientation and distance information.
• Previous research explored texture gradients for inferring 3D scene properties.

Purpose of the Study:

• To reexamine the information content of texture gradients.
• To review relevant psychophysical data on texture perception.
• To introduce a novel texture measure for distance computation.

Main Methods:

• Review of psychophysical studies on texture gradients.
• Introduction of a new texture measure with a defined geometric basis.
• Analysis of variables influencing distance computation accuracy and precision.

Main Results:

• Inferring surface slant from texture gradients is challenging.
• Surface tilt (direction of slant) is reliably determined from texture gradients.
• The new texture measure allows for straightforward distance computation.

Conclusions:

• Texture gradients provide reliable information for surface tilt but not slant.
• A novel, geometrically defined texture measure facilitates accurate distance estimation.
• Understanding factors affecting distance computation is crucial for visual perception models.