Hidden Markov Models for Visual Processing of Marketing Leaflets

The study shows the application of hidden Markov models (HMMs) for the analysis of eye ball movement fixations. The registered visual activity concerns pairwise comparisons of simple advertisement leaflets, differed in their layout orientation and captions’ styles. A simulation experiment was conducted to specify the most appropriate HMMs in terms of information criteria. Six best models were discussed in detail. The identified hidden states together with transition and emission probabilities were the basis of subjects’ visual behavior hypothetical interpretations.