Aesthetics by Numbers: Links between Perceived Texture Qualities and Computed Visual Texture Properties.

Area of Science:

• Visual perception
• Neuroscience
• Psychophysics
• Computational aesthetics

Background:

• Human visual system processes texture for material and environmental assessment.
• Texture perception is crucial for fields like design and art.
• Limited research has explored the link between texture features and high-level judgments.

Purpose of the Study:

• Investigate the relationship between visual texture features and perceived qualities.
• Predict perceived texture qualities, including aesthetic judgments, using computational features.
• Understand the underlying dimensions of human texture judgment.

Main Methods:

• Two experiments involving human observers judging visual textures on various qualities (beauty, roughness, naturalness, etc.).
• Factor analysis to identify key dimensions explaining judgment variability.
• Computation of numerous texture features for each stimulus.
• Correlation analysis between texture features and perceived qualities.

Main Results:

• A two-dimensional space, primarily defined by beauty and roughness, explains approximately 75% of judgment variability.
• Lower spatial frequencies, oblique orientations, higher intensity variation, saturation, and redness correlate with higher beauty.
• Image intensity and uniformity features correlate with roughness ratings.
• Computational texture features predict perceived qualities, including aesthetic appreciation.

Conclusions:

• Human judgment of visual textures relies on a limited set of internal perceptual scales.
• Aesthetic appreciation of textures is predictable from their computed feature content.
• Findings have implications for psychology, neuroscience, design, architecture, and visual arts.