Soft Biometrics; Human Identification Using Comparative Descriptions

• Published in: IEEE transactions on pattern analysis and machine intelligence Vol. 36; no. 6; pp. 1216 - 1228
• Main Authors: Reid, Daniel A., Nixon, Mark S., Stevenage, Sarah V.
• Format: Journal Article
• Language: English
• Published: Los Alamitos, CA IEEE 01.06.2014; IEEE Computer Society; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Applied sciences; Biometric identification; Biometric Identification - methods; Biometrics; Body Size; Bridging; Color; comparisons; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Data processing. List processing. Character string processing; Databases, Factual; Descriptions; Exact sciences and technology; Face; Female; gait biometrics; Human; human descriptions; Humans; Image Processing, Computer-Assisted - methods; Iris recognition; Labels; Male; Materials handling; Memory organisation. Data processing; Pattern analysis; Programming languages; regression; Regression Analysis; Reliability; Representations; retrieval; Semantics; Soft biometrics; Software; Support Vector Machine; Surveillance; Biometrics; Capability index; Gait; human descriptions; Video recording; Posture; Soft biometrics; Programming language; Behavioral analysis; regression; retrieval; Vector support machine; comparisons; gait biometrics
• DOI: 10.1109/TPAMI.2013.219
• ISSN: 0162-8828; 1939-3539

Soft biometrics are a new form of biometric identification which use physical or behavioral traits that can be naturally described by humans. Unlike other biometric approaches, this allows identification based solely on verbal descriptions, bridging the semantic gap between biometrics and human description. To permit soft biometric identification the description must be accurate, yet conventional human descriptions comprising of absolute labels and estimations are often unreliable. A novel method of obtaining human descriptions will be introduced which utilizes comparative categorical labels to describe differences between subjects. This innovative approach has been shown to address many problems associated with absolute categorical labels-most critically, the descriptions contain more objective information and have increased discriminatory capabilities. Relative measurements of the subjects' traits can be inferred from comparative human descriptions using the Elo rating system. The resulting soft biometric signatures have been demonstrated to be robust and allow accurate recognition of subjects. Relative measurements can also be obtained from other forms of human representation. This is demonstrated using a support vector machine to determine relative measurements from gait biometric signatures-allowing retrieval of subjects from video footage by using human comparisons, bridging the semantic gap.