Recognizing Fractal Behavior in Jackson Pollock Artwork through Computer Vision

This project uses Computer Vision to verify that Jackson Pollock's drip-and-pour style paintings exhibit fractal behavior. Pseudocode provided from a different study of fractals in his artwork is converted into a working program that is easily replicable and runnable from a web browser. The Detrended Fluctuation Analysis (DFA) Algorithm is applied to a dataset of Pollock's artwork using Python and Jupytr Notebooks to verify that the output of the algorithm mirrors a power law distribution. The program converts a dataset of Pollock's drip-and-pour style paintings into black-and-white images, and saves off the luminance values into two-dimensional matrices. The algorithm is run on these matrices, and the matrices are then segmented and evaluated at increasingly magnified scales. The final results are then displayed in graphs, which are assessed against a power law distribution graph. The resulting graphs created in this project do verify that fractal behavior is observed. The program is housed in a github repository that can be uploaded and run in Google Colab. The final result is a code base that is broken into multiple sections to help any user gain an understanding of the Computer Vision algorithm.