On Understanding the Search Problem for Image Spaces

Gary R. Greenfield
Bridges: Mathematical Connections in Art, Music, and Science (1999)
Pages 41–54


We consider the problem of analyzing the visual imagery that can be found in a neighborhood of just one image selected from a visual image space. The image space we shall explore is a variant of the space of "expressions" as formulated by Karl Sims. Sims' expressions are rooted trees that can be manipulated and searched using the art-by-choice technique known as "evolving expressions." We present a search methodology and give examples from the catalog of imagery we obtained in searching a neighborhood of a fixed image which contained in excess of 1020 instances of possible images. We draw conclusions about what this portends for future work on search engines developed for exploring artistically non-linear spaces.