Mathematical characterization of Bridget Riley's stripe paintings

I investigate whether mathematical measures can characterize Bridget Riley's stripe paintings. This is motivated by three considerations: (1) stripe paintings are an incredibly constrained art form, therefore it should be relatively straightforward to ascertain whether or not there is a mathematical characterization; (2) Bridget Riley's approach to composition is methodical and thoughtful, so we can assume that her paintings are carefully constructed rather than random and (3) Riley's paintings can appear random on a first glance but have an underlying structure, therefore Riley's works are challenging to characterize because they are close to random while not actually being so. I investigate entropy (both global and local), separation distance and auto-correlation. I find that all can provide some characterization, that entropy provides the best judge between Riley's work and randomly generated variants, and that the entropy measures correlate well with the art-critical descriptions of Riley's development of this style over the five years in which she worked with it. © 2012 Taylor & Francis.