When attempting to code faces for modelling or recognition, estimates of dimensions are typically obtained from an ensemble. These tend to be significantly sub-optimal. Each face contains both predictable and non-predictable qualities; only the predictable aspects are useful for defining coding systems for other faces. Additional information, not coded via the ensemble, is still available. We show that this information can be extracted and described via random Markov fields, and that this can be used to distinguish between images. The distances between images are robust to parameter setting, and can be combined with those derived via ensemble-based techniques to enhance recognition.
Home Contents Author index Keyword index