Related Experiment Videos
Towards a texture naming system: identifying relevant dimensions of texture
1I.B.M., T. J. Watson Research Ctr, Yorktown Heights, NY 10598, USA. rao@watson.ibm.com
Vision Research
|June 1, 1996
Summary
Researchers identified key human texture perception features for data visualization. Three dimensions—repetitive, contrast/directionality, and complexity—were found to be crucial for understanding complex visual data.
Area of Science:
- Computer Vision
- Human-Computer Interaction
- Psychophysics
Background:
- Texture is increasingly used in data visualization to represent multidimensional data.
- Human visual system's sensitivity to texture can overcome display limitations.
- Identifying key textural features is crucial for effective texture perception and visualization.
Purpose of the Study:
- To identify the higher-order textural features perceived by humans.
- To understand how these features are utilized in texture perception.
- To inform the design of texture-based data visualization techniques.
Main Methods:
- An experiment involving 20 subjects rating 56 Brodatz textures on Likert scales.
- Hierarchical cluster analysis and non-parametric multidimensional scaling (MDS) on similarity matrices.
- Classification and Regression Tree Analysis (CART), discriminant analysis, and principal component analysis on rating data.
Main Results:
- MDS plots confirmed the stability of clusters identified by hierarchical analysis.
- MDS solutions showed a good fit to the data (stress=0.12 in 3D).
- Identified three orthogonal dimensions of texture perception: repetitive vs. non-repetitive, contrast/directionality, and complexity (granular/coarse vs. fine).
Conclusions:
- Human perception of texture can be described by three fundamental dimensions.
- These dimensions provide a framework for designing more effective texture-based data visualizations.
- Findings support the use of texture in data visualization by elucidating perceptual underpinnings.