Image-Based Aspect Ratio Selection.

Area of Science:

• Computer Graphics
• Data Visualization
• Image Processing

Background:

• Effective aspect ratio selection is vital for 2D diagrams, but existing methods struggle with general, discrete plots like scatter plots.
• Current techniques often lack perceptual grounding or require time-consuming isoline computations.
• There's a need for efficient and perceptually informed aspect ratio selection for diverse 2D diagram types.

Purpose of the Study:

• To introduce a general, image-based approach for selecting aspect ratios across various 2D diagrams.
• To develop a method that is computationally efficient and perceptually grounded.
• To enhance the characterization of data patterns in visualizations.

Main Methods:

• Derived from Federer's co-area formula and line integral representation for image-based methods.
• Utilizes density fields and bypasses isoline computation for faster processing.
• Employs anisotropic kernel density estimation to accurately represent data patterns.

Main Results:

• The proposed approach offers a faster alternative to existing methods while maintaining perceptual relevance.
• Demonstrated effectiveness through quantitative comparisons and user study validation.
• Successfully applied to diverse diagrams including scatter plots, density plots, and line charts.

Conclusions:

• The image-based method provides a robust and efficient solution for aspect ratio selection in 2D diagrams.
• The approach enhances data pattern characterization, improving visualization clarity.
• Extensions show promise for applications in region of interest (ROI) banking and image data analysis.