Perceptual Similarity: A Texture Challenge

Abstract

Over the last thirty years evaluation of texture analysis algorithms has been dominated by two databases: Brodatz has typically been used to provide single images of approximately 100 texture classes, while CUReT consists of multiple images of 61 physical samples captured under a variety of illumination conditions. While many highly successful approaches have been developed for classification, the challenging question of measuring perceived inter-class texture similarity has rarely been addressed. In this paper we introduce a new texture database which includes a collection of 334 samples together with perceptual similarity data collected from experiments with 30 human observers. We have tested four of the leading texture algorithms and they provide accurate (≈ 100%) performance in a CUReT style classification task, however, a second experiment shows that resulting inter-class distances do not correlate well with the perceptual data.

Color and texture associations in voice-induced synesthesia

“…Textures were uniform with respect to their simulated viewing angle, and presented as gray-scale images to avoid a confounding influence of color. Pictures were taken from the database created by Halley and colleagues (Clarke et al, 2011; Halley, 2011), Brodatz (1966) and from homepages without copyright limitations.…”

“…In addition to color, voice-triggered synesthetes often report texture perceptions, for example a voice might be “smooth but granulated” or “[with a] soft center and very slight fuzziness around the outside.” Despite our knowledge of perceptions such as these, no systematic investigation of visual texture perceptions in synesthesia has yet been conducted—which is perhaps not surprising considering that it is not easy to quantify texture or to relate this quantification to perceptual categories (Petrou et al, 2007; Clarke et al, 2011). Eagleman and Goodale (2009) state: “Quantitatively testing these prevalences [of texture concurrents] will be a challenge: it is straightforward to develop a user-friendly color chooser […], but not so with the multidimensional varieties of texture” (Eagleman and Goodale, 2009, 291).…”

Crowdsourced Feedback With Imagery Rather Than Text

“…Clark et al [1] have shown that in the case of texture images computer vision techniques produce similarity data that does not match human perceptions. Indeed the summarization methods which use metadata and tags are seeking to address the semantic gap between what can be deduced about the meaning of the image from its features and what the image actually means to a viewer.…”

Texture Similarity Estimation Using Contours

“…It is also well-known that humans are extremely adept at exploiting the long-range visual interactions evident in contour information [2,4]. Therefore, we designed an experiment with human observers in order to determine which of three different types of information (2nd-order statistics, local higher order statistics and contour information, see Figure 1) are more important for the perception of texture.Ten human observers were used in a 2AFC (two-alternative forced choice) scheme with 334 texture images drawn from the Pertex database [5]. In each trial the observer was required to compare an original texture image quarter and one variant image quarter ("variant" being one of either contour, power spectrum or randomized block) and decide whether the variant represented the original texture or not (50% of the time they did not).…”

“…Ten human observers were used in a 2AFC (two-alternative forced choice) scheme with 334 texture images drawn from the Pertex database [5]. In each trial the observer was required to compare an original texture image quarter and one variant image quarter ("variant" being one of either contour, power spectrum or randomized block) and decide whether the variant represented the original texture or not (50% of the time they did not).…”