Assessing aesthetics of generated abstract images using correlation structure
This work addresses the problem of aesthetic image generation for researchers in computational creativity and AI art, but it is incremental as it builds on existing methods like compositional pattern-producing networks.
The study investigated whether abstract aesthetic images can be generated without bias from natural or human-selected datasets and found that correlation functions differ for aesthetic images, as confirmed by human subject selections and statistical analysis.
Can we generate abstract aesthetic images without bias from natural or human selected image corpi? Are aesthetic images singled out in their correlation functions? In this paper we give answers to these and more questions. We generate images using compositional pattern-producing networks with random weights and varying architecture. We demonstrate that even with the randomly selected weights the correlation functions remain largely determined by the network architecture. In a controlled experiment, human subjects picked aesthetic images out of a large dataset of all generated images. Statistical analysis reveals that the correlation function is indeed different for aesthetic images.