Identifying humans using comparative descriptions

Soft biometrics is a new form of biometric identification which utilizes human descriptions of a subject´s physical appearance. Although these descriptions intuitively have less discriminatory capability than traditional biometric approaches, they are able to retrieve and recognize subjects based solely on a human description. To permit soft biometric identification the human description must be accurate, yet conventional human descriptions comprising of absolute labels and estimations are often unreliable. In this paper we introduce a novel method of human description which utilizes comparative descriptors derived from visual comparisons between subjects. This innovative approach to obtaining human descriptions has been shown to counter many problems associated with absolute categorical labels. Comparative categorical labels are objective and can be used to infer descriptive continuous relative measurements. The resulting biometric signatures have been demonstrated to differ significantly from absolute descriptions allowing improved retrieval of subjects and could even be used to increase the accuracy of witness description in crime analysis.