READ ME

Project Title: Image Decomposer

Author

• Name: Izabela Maria Sztuka

• Affiliation: Center for Environmental Neuroscience, Max Planck Institute for Human Development

• Contact: [email protected]/[email protected]

• Version: 4.0

• Data of last update: 07/24

Publication

• Sztuka, I. M., & Kühn, S. Blue Skies: Does Visual Composition of Sky Guide Subjective Judgments of Naturalness in the Environment?. Available at SSRN: https://ssrn.com/abstract=4741220 or http://dx.doi.org/10.2139/ssrn.4741220
• Sztuka, I. M., Becker, M. & Kühn, S. Neural representations underlying psychological responses to natural and artificial features in indoor architecture. (in review, Journal for Environmental Psychology)

Table of Contents

• Overview
• Repository Structure
• Prerequisites
• Setup
• Usage
• Notes
• Acknowledgments

ImageDecomposer

Overview

The ImageDecomposer is a MATLAB class that decomposes images into various features using multiple toolboxes. This project is designed to handle large batches of images, extract their features, and save the results in a CSV file. It uses a combination of several toolboxes, including Berman, CircStat, Str_Nstr_DecomposerV3, Count_Colors_variableDupIMS, boxcount, and LGNstatistics.

Repository Structure

├──ImageDecomposer.m # MATLAB class for image decomposition ├── mainScript.m # Main script to run the decomposition ├── toolbox/ # Directory containing various toolboxes ├── stimuli/ # Directory containing input images ├── output/ # Directory to store output CSV file ├── load_featureheaders.csv # CSV file with feature headers └── README.md # Notes and instructions

Prerequisites

• MATLAB installed on your system (tested with 2019 up)
• Required toolboxes placed in the toolbox directory:
  • WholeIm_Decomposer_cc_ims.m (Berman toolbox)
  • CircStat2012a (CircStat toolbox)
  • Str_Nstr_DecomposerV3ci (Str_Nstr_DecomposerV3 toolbox)
  • Count_Colors_variableDupIMS.m
  • boxcount.m
  • LGNstatistics-master
• Note these toolboxes have been modified first by Carlos then by me to work with this script.

Setup

1. Clone this repository to your local machine.
2. Ensure that all required toolboxes are available in the toolbox directory.
3. Place your input images in the stimuli directory.

Usage

1. Open MATLAB and navigate to the repository directory.
2. Open mainScript.m and modify parameters of the analysis.
3. Run the mainScript.m to start the image decomposition process.

Process

• The script performs the following steps:

1. Sets up the paths and directories.
2. Initializes the ImageDecomposer class.
3. Processes each image in the stimuli directory.
4. Loads feature names from load_featureheaders.csv.
5. Reads the decomposed data from output/stimuli.csv and updates the feature names.
6. Saves the updated data back to output/stimuli.csv.

Example

% Set up paths and parameters
root = pwd;
image_folder = "stimuli";
output = fullfile(root, 'output');

% Initialize the ImageDecomposer class
decomposer = ImageDecomposer(root, output, image_folder);

% Process the images
decomposer.decomposeImages();

Notes

Contents

• WholeIm_Decomposer - toolbox to extraxt hue, saturation, brightness, entropy, edge density. Modified in LMG by CC & IMS. Method used in Nature_gradient & ARCH studies. Published: [1] Kardan, O., Demiralp, E., Hout, M. C., Hunter, M. R., Karimi, H., Hanayik, T., ... & Berman, M. G. (2015). Is the preference of natural versus man-made scenes driven by bottom�up processing of the visual features of nature?. Frontiers in psychology, 6. [2] Berman, M. G., Hout, M. C., Kardan, O., Hunter, M. R., Yourganov, G., Henderson, J. M., ... & Jonides, J. (2014). The perception of naturalness correlates with low-level visual features of environmental scenes. PloS one, 9(12), e114572.

• CircStat2012a - auxiliary toolbox for circular statistics. Used by WholeIm_Decomposer. Published: [3] Berens, P., CircStat: A Matlab Toolbox for Circular Statistics, Journal of Statistical Software, Volume 31, Issue 10, 2009

• SHINEtoolbox - toolbox to control for low-level image properties by normalising contrast and luminance. Published: [4] Willenbockel, V., Sadr, J., Fiset, D., Horne, G. O., Gosselin, F., & Tanaka, J. W. (2010). Controlling low-level image properties: the SHINE toolbox. Behavior research methods, 42(3), 671-684.

• Str_Nstr_DecomposerV3 - 3 versions of function to extract straight & non-straight edge density. Used in publications [1] & [2]. Modified by CC (Str_NStr_DecomposerV3c.m) and IMS (Str_NStr_DecomposerV3ci.m). Method used in Nature_gradient & ARCH studies.

• Count_Colors_variableDup - function to qualtify colours on the images. Created by Carlos Cassanello.

• fractal_analysis - function conducting fractal analysis of the image using boxcounting with either binary or differential method. Created by IMS. Method as used in: [5] Nature_gradient (https://doi.org/10.3389/fpsyg.2022.932507) & ARCH studies (in preparation)

Acknowledgments