Conjoint analysis to measure the perceived quality in volume rendering

Area of Science:

• Computer Graphics
• Human-Computer Interaction
• Perception Science

Background:

• Visualization algorithms often possess numerous parameters, creating high-dimensional spaces for rendering results.
• Systematic analysis of perceived quality is crucial for optimizing these parameters, but exhaustive testing is infeasible.
• Parameter co-dependencies further complicate the optimization process.

Purpose of the Study:

• To introduce an efficient user study design and analysis strategy for optimizing visualization parameters.
• To address the challenge of high-dimensional parameter spaces in visualization.
• To enable fast and easy acquisition of user feedback without exhaustive parameter testing.

Main Methods:

• Modified conjoint analysis, a preference measuring methodology from psychology and market research.
• Developed an efficient framework for user study design and analysis.
• Applied the framework to measure perceived quality in volume rendering with large parameter spaces.

Main Results:

• Demonstrated an efficient method for obtaining user feedback on visualization quality.
• Successfully measured perceived quality within complex, large parameter spaces.
• The conjoint analysis approach proved effective for preference measurement in this context.

Conclusions:

• The proposed user study design and analysis strategy efficiently handles high-dimensional parameter spaces in visualization.
• Conjoint analysis offers a viable alternative to exhaustive testing for optimizing visualization parameters based on perceived quality.
• This framework facilitates faster and more practical optimization of visualization algorithms.