Motion coding for pattern detection

Abstract

How people perceive motion has been the subject of study of vision researchers for decades and it is well known that humans are highly sensitive to motion patterns (for a recent review see Blake and Lee, 2005]. Of particular interest is the finding that motion is processed by different subsystems in the brain than color [e.g. Livingstone and Hubel, 1988] suggesting that there may be relatively little interference between data displayed using motion and other data displayed using color. AbstractA relatively underutilized method for visualizing data is to map variables directly to the oscillatory motion of glyphs. When doing this, the most straightforward parameters to consider are the frequency, phase and amplitude of sinusoidal oscillation. We report the results of an experiment that used a staircase procedure to assess human sensitivity to the frequency, phase and amplitude of motion as a method for revealing two-dimensional spatial patterns in data. For comparison, we displayed the same targets using glyph color on a red-green scale and glyph value on a grey scale. Both large and small glyphs were used. Our results show that subjects were most sensitive to spatial patterns mapped to relative motion phase. Subjects were least sensitive to the frequency of oscillation. Grey scale and color mapping were ineffective when the glyphs were small but somewhat effective with larger glyphs. Various issues concerning the use of motion for data display are discussed.If motion is to be used for information display it generally makes the most sense for that motion to be anchored, oscillating about a fixed point [Bartram et al, 2003], because under normal circumstances the position of that point also conveys important information.A mathematically simple form of anchored motion is sinusoidal oscillation. As illustrated in Figure 1 the oscillation of a data point can be varied in its frequency, phase and amplitude of motion. It is also possible for such motion to be varied independently in both the x and y directions. But this does not mean that six parameters are available for data mapping because points moving at different frequencies will constantly be moving in and out of phase with one another. Thus effectively we can either vary the frequency and amplitude of motion, or the phase and amplitude of motion. We can do this in both the x and y directions, yielding four display parameters for simple sinusoidal motion on a plane. The study we present here, however, only mapped a single data variable to motion, although it redundantly resulted in both x and y motion.

Splatting the Lines in Parallel Coordinates

“…where N g and N total represent the number of items in cluster g and the total number of items in the dataset respectively. Ware and Bobrow [WB06] found that pattern detection in motion is most effective when all the glyphs are moving at the same frequency but in different phases. Based on their suggestions, we use the dynamically‐changed speed to control the phase differences between clusters and reduce crossings between lines from different clusters.…”

Using Visual Design Experts in Critique-Based Evaluation of 2D Vector Visualization Methods

“…Among the perceptually based cues proposed are preattentive visual cues such as color and orientation [7] and Gestalt principles of grouping according to proximity, similarity, or closure; even motion has been used to disambiguate visual elements [23] or to animate time-varying data.…”

Visual Multiplexing

“…color weaving with blending for multifield visualization [HKIH07]. Ware studied glyph‐based texture in [War09], and oscillatory motion of glyphs in [WB06]. Figure 3(h) shows an example of a texture pattern made of a collection of simple glyphs, called textons [War09] (cf.…”