Perception and Quantization Model for Periodic Contour Modifications

Dmitri Presnov, Andreas KolbJournal of Imaging 2022 · DOI

TL;DR

This paper introduces a perceptual model for understanding how humans perceive periodic modifications to shapes' contours, used in data visualization. By conducting a user study, they've developed a way to map visual stimuli (like wave amplitude and frequency) to perceived magnitudes, enabling the creation of more perceptually uniform and legible visualizations.

What Problem Does This Paper Address?

In data visualization, we often use visual cues to represent information. One technique involves modifying the contours of shapes (like circles or icons) with periodic waves. These waves can encode data through their amplitude (how "big" the wave is) and frequency (how many waves there are). While this approach is promising for adding more visual channels to visualizations, there's been a lack of systematic research into how humans actually perceive these contour modifications. This means designers might not be creating visualizations that are truly "perceptually uniform" – where equal changes in data lead to equal perceived changes – or that are easily legible. This paper aims to fill that gap by investigating the perception of these periodic contour modifications, both geometric and color-based, and developing a model to guide their use in visualization design.

Methods

The core of this research is a user study designed to understand how people perceive periodic contour modifications. They explored two main types of modifications:

  • Geometric Modifications: Altering the actual shape of the contour using waves with varying amplitude, frequency, and waveform (sinusoidal, rectangular, sawtooth).
  • Color Modifications: Alternating segments of the contour's color (intensity) to create a periodic effect, also varying in amplitude and frequency.

The study involved several types of experiments:

  • Magnitude Estimation: Participants were shown reference shapes with known stimulus values and asked to estimate the perceived magnitude of a test shape. This helps establish the relationship between what's presented visually and what's perceived.
  • Waveform Calibration: This experiment aimed to see how much different waveforms (rectangular, sawtooth) affected perception compared to a standard sinusoidal wave.
  • Perceptual Dependencies: They investigated how changes in one visual variable (e.g., amplitude) affected the perception of another (e.g., frequency).

Based on the user study data, they developed:

  • Stimulus-to-Perception Transformation Functions: These are mathematical models (power functions) that describe how stimulus magnitudes translate into perceived magnitudes.
  • Quantization Scheme: This defines perceptually equidistant levels for each visual variable, ensuring that equally spaced data points are perceived as such.
  • Waveform Calibration Model: A method to adjust perception for different waveforms.
  • Size Transfer Method: A proposed approach to apply the developed quantization to shapes of different sizes.

Datasets

The paper does not mention using pre-existing datasets. Instead, they generated their own stimuli for the user study. These stimuli were created from four monochrome base shapes, modified with geometric and color-based periodic contour variations. The specific ranges and steps for amplitude and frequency were carefully chosen based on prior research and to avoid strong perceptual interferences.

Key Insights and Ideas

  • Linearity for Amplitude, Quadratic for Frequency: The perception of both geometric and color amplitude follows a largely linear relationship with the stimulus. However, perceived frequency shows a closer-to-quadratic relationship with the stimulus. This is a crucial finding for designing scales.
  • Waveform Influence is Marginal: The specific shape of the wave (sinusoidal, rectangular, sawtooth) has a surprisingly small impact on the perception of amplitude and frequency. This simplifies visualization design, as designers don't need to worry too much about waveform choice for these variables.
  • Frequency is More Discriminable: Both geometric and color frequency can be distinguished into more levels than their respective amplitudes. This suggests frequency is a more potent visual variable for encoding data.
  • Perceptual Uniformity is Achievable: By understanding these perceptual relationships, it's possible to create quantization levels that are visually equidistant, leading to more accurate data representation.
  • Transferring to Different Sizes is Possible (with caveats): The authors propose a method to adapt their fixed-size model to different glyph sizes, though they acknowledge this requires further validation.

Key Results

  • Stimulus-to-Perception Models: They derived power functions (e.g., e(gas) = 0.9097* ga^1.0604 -0.1814 for geometric amplitude) that model the relationship between stimulus and perception.
  • Quantization Levels: The study identified a specific number of distinguishable, perceptually equidistant levels for each visual variable. For example, geometric amplitude has 4 levels, geometric frequency has 5, and both color amplitude and frequency have 4 levels within the tested ranges.
  • Waveform Calibration: The calibration functions showed that the influence of waveform is "rather marginal," with scaling factors close to 1.
  • Amplitude-Frequency Dependencies: While some dependencies were observed, they were generally "rather marginal" and could likely be neglected in visualization design.

Limitations and Future Work

  • Limited Scope of Waveforms and Shapes: The study focused on sinusoidal, rectangular, and sawtooth waveforms and four base shapes. Future work could explore a wider variety.
  • Fixed Stimulus Ranges: The study deliberately narrowed the ranges of geometric variables to avoid strong interferences. Further investigation is needed to determine the impact of critical magnitudes and how these limits might vary with different base shapes.
  • Color Modifications: The current study was limited to black-and-white images. Extending this to colored shapes and exploring potential reductions in color amplitude levels based on base shape intensity is a future direction.
  • Combined Modifications: The paper only considered individual modifications. Investigating the perceptual impact of combining geometric and color modifications is a significant area for future research.
  • Size Transfer Validation: The proposed method for transferring the model to different sizes is heuristic and requires experimental validation.
  • Direct JND Tests: The current quantization is statistically estimated. Comparing these results with direct Just Noticeable Difference (JND) tests could provide further insights.

Summary updated

Abstract

Periodic, wave-like modifications of 2D shape contours are often applied to convey quantitative data via images. However, to the best of our knowledge, there has been no in-depth investigation of the perceptual uniformity and legibility of these kind of approaches. In this paper, we design and perform a user study to evaluate the perception of periodic contour modifications related to their geometry and colour. Based on the study results, we statistically derive a perceptual model, which demonstrates a mainly linear stimulus-to-perception relationship for geometric and colour amplitude and a close-to-quadratic relationship for the respective frequencies, with a rather negligible dependency on the waveform. Furthermore, analyzing the distribution of perceived magnitudes and the overlapping of the respective 50% confidence intervals, we extract distinguishable, visually equidistant quantization levels for each contour-related visual variable. Moreover, we give first insights into the perceptual dependency between amplitude and frequency, and propose a scheme for transferring our model to glyphs with different size, which preserves the distinguishability and the visual equidistance. This work is seen as a first step towards a comprehensive understanding of the perception of periodic contour modifications in image-based visualizations.
BibTeX
@article{presnov2022perception,
  author = {Dmitri Presnov and Andreas Kolb},
  title = {{Perception and Quantization Model for Periodic Contour Modifications}},
  journal = {Journal of Imaging},
  year = {2022},
  doi = {10.3390/jimaging8110311},
}
Figure