Statistical image properties of print advertisements, visual artworks and images of architecture

Statistical image properties of print advertisements, visual artworks and images of architecture Julia Braun Seyed A. Amirshahi Joachim Denzler Christoph Redies

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Statistical image properties of print advertisements, visual artworks and images of architecture 2013 10.3389/fpsyg.2013.00808 statistical-image-properties-of-print-advertisements statistical-image-properties-of-print-advertisements.txt, the project's staged copy 796c92452f47aa4d0fcc040659c825813487d977 cc-by empirical-aesthetics

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them higher-order (global) statistical image properties of print advertisements, visual artworks and images of architecture, measured with computer-vision algorithms Quotation withheld because this record is not cleared for redistribution. whether images of advertisements (and architecture) share the statistical image properties characteristic of artworks - high self-similarity, intermediate complexity, low anisotropy - and whether the measures discriminate image categories Quotation withheld because this record is not cleared for redistribution. the first analysis of higher-order global statistical image properties of print advertisements, showing that man-made visually pleasing images share specific higher-order properties that distinguish them from other image types Quotation withheld because this record is not cleared for redistribution. existence Quotation withheld because this record is not cleared for redistribution. man-made visually pleasing images (artworks, advertisements, architecture) share a specific pattern of higher-order statistical image properties that distinguishes them from other types of images Quotation withheld because this record is not cleared for redistribution. corpus-analysis corpus-analysis Quotation withheld because this record is not cleared for redistribution. The human visual system is adapted to process natural-scene statistics efficiently, so visually pleasing images that follow universal regularities can be processed efficiently. efficient visual processing of natural-scene statistics Quotation withheld because this record is not cleared for redistribution. It has been proposed that stimuli which can be processed fluently are more aesthetic in general. processing fluency Quotation withheld because this record is not cleared for redistribution. Higher-order image features may be processed fast at a lower level of the visual system, similar to gist perception of scenes, possibly before advertisement content is recognised. fast gist-level processing Quotation withheld because this record is not cleared for redistribution. Advertisements can activate the brain's reward circuitry, which is also involved in the perception of other visually pleasing images such as artworks. reward-circuitry activation Quotation withheld because this record is not cleared for redistribution. Berlyne's u-shaped dependence of hedonic value on complexity was explained by the differing activation of two antagonistic brain systems, a reward system and an aversion system. Berlyne two-system account of complexity Quotation withheld because this record is not cleared for redistribution. Humans may prefer fractal, self-similar geometry in architecture and urban scenes, possibly because ancestral natural surroundings had fractal characteristics. biophilic preference for fractal geometry Quotation withheld because this record is not cleared for redistribution. Higher anisotropy in advertisements may arise from vertical text and image divisions along cardinal (horizontal and vertical) orientations. anisotropy from cardinal orientations Quotation withheld because this record is not cleared for redistribution. Whether these image properties are necessary or sufficient to induce aesthetic perception, and how they correlate with brain activation, remains to be investigated. Quotation withheld because this record is not cleared for redistribution. The measured properties are not independent, so the precise relation between the different measures remains to be determined. Quotation withheld because this record is not cleared for redistribution. The dependence of beauty on complexity is not straightforward, and there is a general lack of well-controlled studies that manipulate complexity in such images. Quotation withheld because this record is not cleared for redistribution. Monochrome artworks are not equivalent to grayscale versions of colored paintings, because conversion to grayscale may destroy their aesthetic appeal. Quotation withheld because this record is not cleared for redistribution. The role of anisotropy in architectural aesthetics remains unclear. Quotation withheld because this record is not cleared for redistribution. Whether the specific pattern of the three measures is required or sufficient for aesthetic perception remains to be studied. Quotation withheld because this record is not cleared for redistribution. statistical image properties statistical image properties Quotation withheld because this record is not cleared for redistribution. higher-order image properties higher-order image properties Quotation withheld because this record is not cleared for redistribution. self-similarity self-similarity Quotation withheld because this record is not cleared for redistribution. complexity complexity Quotation withheld because this record is not cleared for redistribution. anisotropy anisotropy Quotation withheld because this record is not cleared for redistribution. Birkhoff-like measure Birkhoff-like measure Quotation withheld because this record is not cleared for redistribution. fractal dimension fractal dimension Quotation withheld because this record is not cleared for redistribution. Fourier slope Fourier slope Quotation withheld because this record is not cleared for redistribution. PHOG method PHOG method Quotation withheld because this record is not cleared for redistribution. gradient image gradient image Quotation withheld because this record is not cleared for redistribution. box-counting method box-counting method Quotation withheld because this record is not cleared for redistribution. aesthetic perception aesthetic perception Quotation withheld because this record is not cleared for redistribution. experimental aesthetics experimental aesthetics Quotation withheld because this record is not cleared for redistribution. reward circuitry reward circuitry Quotation withheld because this record is not cleared for redistribution. visually pleasing images visually pleasing images Quotation withheld because this record is not cleared for redistribution. man-made images man-made images Quotation withheld because this record is not cleared for redistribution. advertisements advertisements Quotation withheld because this record is not cleared for redistribution. artworks artworks Quotation withheld because this record is not cleared for redistribution. architecture architecture Quotation withheld because this record is not cleared for redistribution. single-product advertisements single-product advertisements Quotation withheld because this record is not cleared for redistribution. leaflets leaflets Quotation withheld because this record is not cleared for redistribution. urban scenes urban scenes Quotation withheld because this record is not cleared for redistribution. buildings buildings Quotation withheld because this record is not cleared for redistribution. natural scenes natural scenes Quotation withheld because this record is not cleared for redistribution. aesthetic value aesthetic value Quotation withheld because this record is not cleared for redistribution. order order Quotation withheld because this record is not cleared for redistribution. biophilia biophilia Quotation withheld because this record is not cleared for redistribution. higher-order image properties → aesthetic perception The paper speculates that specific higher-order image properties may enhance visual effectiveness and play a causative role in judging stimuli pleasing, but whether they do remains to be investigated. — the relation stated: higher-order image properties may play a causative role in perceiving stimuli as pleasing Quotation withheld because this record is not cleared for redistribution. processing fluency → aesthetic value Stimuli that can be processed fluently are proposed to be more aesthetic in general. — the relation stated: fluent processing is proposed to increase aesthetic response Quotation withheld because this record is not cleared for redistribution. complexity → fractal dimension Complexity and the fractal dimension are the most strongly correlated of the measures studied. — the relation stated: complexity is strongly correlated with the fractal dimension Quotation withheld because this record is not cleared for redistribution. self-similarity → complexity Self-similarity and complexity are also relatively strongly correlated across all images. — the relation stated: self-similarity correlates with complexity Quotation withheld because this record is not cleared for redistribution. self-similarity → PHOG method Self-similarity, complexity and anisotropy are the measures obtained with the PHOG method. — the relation stated: self-similarity is one of the measures Quotation withheld because this record is not cleared for redistribution. complexity - measured sense: the mean of all gradient strengths in the gradient image, an objective image statistic Quotation withheld because this record is not cleared for redistribution. complexity - hedonic/conceptual sense: Berlyne's information content, whose intermediate level is judged more aesthetic Quotation withheld because this record is not cleared for redistribution. complexity - fractal-dimension sense: the density of edges in a binarized image Quotation withheld because this record is not cleared for redistribution. self-similarity - measured sense: PHOG histogram-intersection similarity of image sections to a reference section across pyramid levels Quotation withheld because this record is not cleared for redistribution. self-similarity - fractal/aesthetic sense: fractal-like self-similar structure that can evoke aesthetic experience Quotation withheld because this record is not cleared for redistribution. anisotropy - measured sense: statistical variance of image-feature strength across orientations, distinguishing artworks (low) from advertisements (higher) Quotation withheld because this record is not cleared for redistribution. aesthetic value - Birkhoff sense: a function of the ratio of order (here self-similarity) to complexity Quotation withheld because this record is not cleared for redistribution. aesthetics - computational sense: statistical image properties relatable to visual perception, as against the perceptual experience of pleasingness Quotation withheld because this record is not cleared for redistribution. builds-on → the computational/statistical-image-properties approach to visual aesthetics (Redies, Amirshahi, Graham and colleagues) note: Applies previously developed PHOG measures and Fourier/fractal analyses to advertisements and architecture. Quotation withheld because this record is not cleared for redistribution. extends → prior PHOG analyses that distinguish artworks from other image categories note: Adds advertisements and architecture to the categories distinguished by the PHOG measures. Quotation withheld because this record is not cleared for redistribution. supports → the idea that visually pleasing man-made images share specific higher-order statistical properties allowing efficient visual processing note: Speculative extension of the idea to print advertising. Quotation withheld because this record is not cleared for redistribution. supports → the biophilia hypothesis that humans prefer fractal, self-similar geometry in architecture note: Self-similarity and complexity similarities support it, but anisotropy is much higher than in natural scenes. Quotation withheld because this record is not cleared for redistribution. qualifies → Berlyne's link between intermediate complexity and aesthetic preference note: Support is qualified by the lack of well-controlled complexity-manipulation studies. Quotation withheld because this record is not cleared for redistribution. contrasts-with → prior studies of basic print-advertisement features (color, size, spacing) note: This work targets higher-order global properties not previously studied. Quotation withheld because this record is not cleared for redistribution. advertisements —is-a→ man-made images advertisements man-made images the relation itself is separately evidenced · strength of the relation: 1 · note: Flat categorical kindhood, cue 'another type of'. · the relation stated: Advertisements are another type of man-made images that attract human attention. Quotation withheld because this record is not cleared for redistribution. complexity —operationalised-as→ gradient image complexity gradient image the relation itself is separately evidenced · strength of the relation: 0.95 · note: Measurement/definition of complexity via the gradient image. · the relation stated: Complexity is operationalised as the mean of all gradient strengths in the gradient image. Quotation withheld because this record is not cleared for redistribution. fractal dimension —operationalised-as→ box-counting method fractal dimension box-counting method the relation itself is separately evidenced · strength of the relation: 0.95 · note: Fractal dimension operationalised via the box-counting method. · the relation stated: The fractal (box-counting) dimension is estimated by fitting a line to the box-counting plot. Quotation withheld because this record is not cleared for redistribution. advertisements —contrasted-with→ artworks advertisements artworks the relation itself is separately evidenced · strength of the relation: 0.85 · note: Differentiator: anisotropy; the two share most other properties. · the relation stated: Median anisotropy is higher for advertisements than for artworks, which are distinctively low in anisotropy. Quotation withheld because this record is not cleared for redistribution. single-product advertisements —contrasted-with→ leaflets single-product advertisements leaflets the relation itself is separately evidenced · strength of the relation: 0.9 · note: Differentiator: complexity. · the relation stated: Single-product advertisements are generally less complex than leaflets that promote multiple products per page. Quotation withheld because this record is not cleared for redistribution. urban scenes —contrasted-with→ buildings urban scenes buildings the relation itself is separately evidenced · strength of the relation: 0.9 · note: Differentiator: self-similarity. · the relation stated: In contrast to buildings and facades, urban scenes contain more diverse elements and are less self-similar. Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span is about buildings' self-similarity; 'urban scenes' is loose. Quotation withheld because this record is not cleared for redistribution. On average, images of advertisements and artworks share a similar degree of complexity (fractal dimension) and self-similarity and similarities in the Fourier spectrum, but advertisements are more anisotropic than artworks. Quotation withheld because this record is not cleared for redistribution. Images of advertisements are characterized by a specific combination of higher-order image properties: high self-similarity, intermediate complexity and intermediate anisotropy. Quotation withheld because this record is not cleared for redistribution. For single-product advertisements, the measured image properties do not depend on the type of product promoted (cars, cosmetics, fashion or other). Quotation withheld because this record is not cleared for redistribution. Median anisotropy is higher for advertisements than for artworks and natural scenes, and nearly all categories are significantly more anisotropic than artworks except the natural categories. Quotation withheld because this record is not cleared for redistribution. The measured properties are not independent; the strongest correlation is between complexity and the fractal dimension. Quotation withheld because this record is not cleared for redistribution. The PHOG measures (self-similarity, complexity and anisotropy) allow distinguishing artworks from many other image categories on average. Quotation withheld because this record is not cleared for redistribution. Statistical testing showed overall differences between all 15 image groups on all six measures (p < 0.001). Quotation withheld because this record is not cleared for redistribution. Images of architecture share statistical properties with images of advertisements to a large extent. Quotation withheld because this record is not cleared for redistribution. The three PHOG approaches to self-similarity (parent, neighbor, ground) give different absolute values but are robust for the relative differences between image categories. Quotation withheld because this record is not cleared for redistribution. the statistical variance (heterogeneity) of image-feature strength across orientations in an image; low anisotropy means uniform strengths across orientations, high means orientations differ in prominence anisotropy Quotation withheld because this record is not cleared for redistribution. the mean of all gradient strengths in the gradient image; low for uniform images, high for images with large pixel changes complexity Quotation withheld because this record is not cleared for redistribution. the density of edges in a binarized image, an indicator of complexity closely related to it fractal dimension Quotation withheld because this record is not cleared for redistribution. the degree to which image sections resemble a reference section - the parent section, the neighboring sections, or the whole image at ground level - across pyramid levels, computed with the PHOG method self-similarity Quotation withheld because this record is not cleared for redistribution. an aesthetic-value proxy defined as self-similarity divided by complexity, substituting self-similarity for order in Birkhoff's order/complexity ratio Birkhoff-like measure Quotation withheld because this record is not cleared for redistribution. the concept that urban planning and architectural design should be based on fractal (self-similar) geometry, which humans are held to prefer in their environment biophilia Quotation withheld because this record is not cleared for redistribution. the slope of the log-log plot of the radially averaged Fourier power spectrum; about -2 for scale-invariant, fractal-like monochrome artworks and natural scenes Fourier slope Quotation withheld because this record is not cleared for redistribution. about 200 color images per dataset — size of each of the 15 image-category datasets Quotation withheld because this record is not cleared for redistribution. 15 image categories — number of image categories analysed Quotation withheld because this record is not cleared for redistribution. 197 paintings — size of the artworks dataset Quotation withheld because this record is not cleared for redistribution. self-similarity 0.62 (advertisements) vs 0.68 (artworks), ground approach — PHOG self-similarity, advertisements compared to artworks Quotation withheld because this record is not cleared for redistribution. complexity 9.00 (advertisements), 7.23 (artworks), 28.25 (branches) — PHOG complexity values across categories Quotation withheld because this record is not cleared for redistribution. anisotropy 0.70x10-3 (advertisements) vs 0.49x10-3 (artworks) — anisotropy is higher for advertisements than artworks Quotation withheld because this record is not cleared for redistribution. Fourier slope -2.57 (advertisements), -2.45 to -2.53 (architecture), -2.77 (artworks) — slope of the radially averaged Fourier power spectrum by category Quotation withheld because this record is not cleared for redistribution. 100,000 pixels — image size to which all images were reduced for the PHOG analysis Quotation withheld because this record is not cleared for redistribution. 16 bins covering 360 degrees — orientation bins used to compute the HOG features Quotation withheld because this record is not cleared for redistribution. level 3 (64 sub-sections) — highest pyramid level used in the analysis Quotation withheld because this record is not cleared for redistribution. self-similarity self-similarity Quotation withheld because this record is not cleared for redistribution. complexity complexity Quotation withheld because this record is not cleared for redistribution. anisotropy anisotropy Quotation withheld because this record is not cleared for redistribution. fractal dimension fractal dimension Quotation withheld because this record is not cleared for redistribution. Fourier slope Fourier slope Quotation withheld because this record is not cleared for redistribution. Birkhoff-like measure Birkhoff-like measure Quotation withheld because this record is not cleared for redistribution. image category image category Quotation withheld because this record is not cleared for redistribution. type of product type of product Quotation withheld because this record is not cleared for redistribution. resolution of the images resolution of the images Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.3389/fpsyg.2013.00808.