Image statistics substantiate Gaudí's naturalistic design principles
Data files
May 20, 2025 version files 279.80 MB
-
Data.csv
4.40 KB
-
get_alpha.m
1.84 KB
-
image_analysis.m
893 B
-
README.md
3.08 KB
-
statistical_analysis2024.prism
279.79 MB
Abstract
Human observers perceive natural scenes differently from man-made environments, and this distinction can be quantified using image statistics. However, there is limited evidence on how the architectural style of buildings influences these statistics and, consequently, visual perception. Understanding this relationship is crucial, as architectural design shapes our visual and psychological experience of built environments. The amplitude spectrum slope reflects the sharpness and detail of the image. This measure tends to be closer to 1 among photographs of natural scenes compared to those of man-made objects. Additionally, photographs with higher entropy, indicating increased unpredictability and information, are more likely to capture attention. In the present study, we examined photographs of buildings designed by Antoni Gaudí (1852-1926), known for his nature-inspired creations. Our analysis reveals that photographs of Gaudí's buildings exhibited an average amplitude spectrum slope more aligned with natural scenes than contemporary structures, along with higher image entropy, indicating greater visual complexity. Notably, the presence of trees in front of contemporary buildings enhances their naturalistic qualities, reflected in shifts in amplitude spectrum slope and entropy toward values characteristic of natural scenes. These findings suggest that integrating naturalistic design elements into architecture can modify image statistics, potentially influencing perception and aesthetic experience.
Authors: Olga Dyakova, Karin Nordström, Christian Benedict
[Access this dataset on Dryad](Dataset DOI link)
Description of the data and file structure
Below is a list of files, including raw images and analysis, associated with the paper.
.csv - Data is provided in Excel format and can be viewed or edited using programs such as LibreOffice Calc, OpenOffice Calc, Microsoft Excel, or imported into Google Sheets.
.m - MATLAB scripts for image analysis.
.prism GraphPad Prism files for statistical analysis, viewable and editable with GraphPad Prism software.
Description of the data and file structure
Data.csv
This Excel file contains data for future analysis:
- The column entropy: Gaudí's buildings contains entropy values for 27 images of Gaudí's buildings. The entropy was extracted using MATLAB.
- The column entropy: contemporary buildings contains entropy values for 29 images of contemporary buildings. The entropy was extracted using MATLAB.
- The column entropy: natural scenes contains entropy values for 60 images of natural scenes. The entropy was extracted using MATLAB.
- The column amplitude spectrum slope: Gaudí's buildings contains amplitude spectrum slope values for 27 images of Gaudí's buildings. The entropy was extracted using MATLAB.
- The column amplitude spectrum slope: contemporary buildings contains amplitude spectrum slope values for 29 images of contemporary buildings. The entropy was extracted using MATLAB.
- The column amplitude spectrum slope: natural scenes contains amplitude spectrum slope values for 61 images of natural scenes. The entropy was extracted using MATLAB.
The amplitude spectrum slope measures the sharpness and level of detail in an image. To compute the average amplitude spectrum slope, we first applied a Fourier transform to the greyscale images. Next, we calculated the average amplitude across all orientations as a function of spatial frequency and plotted the results on a log–log scale. A linear function was then fitted to the amplitude spectrum (following the form 1/f^α) within the range of 10 to 100 cycles per image (cpI). The slope of this fit, representing the amplitude spectrum slope, was extracted and rounded to two decimal places.
Image entropy quantifies the randomness or information content within an image. We calculated the Shannon entropy using Matlab’s function 'entropy'.
Statistical_analysis.prism
This GraphPad Prism file contains statistical analysis and figures.
image_analysis.m.
This MATLAB file contains the function to calculate the amplitude spectrum and the amplitude slope constant.
This is the same code described in Dyakova et al. (2015), A higher order visual neuron tuned to the spatial amplitude spectra of natural scenes, Nature Communications, 6, 8522 (https://www.nature.com/articles/ncomms9522).
We only added a line to extract the amplitude spectrum as output.
Please read associated README file
- Dyakova, Olga; Nordström, Karin; Benedict, Christian (2025). Image statistics substantiate Gaudí’s naturalistic design principles. Scientific Reports. https://doi.org/10.1038/s41598-025-06007-8
