Following on from the development of facial recognition, biometric research has now taken a further step forward with the creation of a recognition system based on a person’s silhouette and individual way of walking.

Silhouette and Gait Analysis Enhances the Biometric Toolkit

Ramon Mollineda, a senior lecturer at the Department of Computing Languages and Systems at the Jaume I University near Valencia in Spain, is currently working on the development of a new biometric technique based on the way a person walks and his/her silhouette. The technique offers a significant advantage over other biometric methods in that it enables recognition of a person at a distance, thus avoiding the need for the subject’s cooperation. Detecting suspicious behaviour via video surveillance, monitoring access to buildings and restricted areas, and analysing a specific population in terms of gender and age are among the potential applications of the new technology.

Getting around the need for consent

Every person has a very individual way of walking, and while it is true that we can all make a conscious effort to alter our habitual gait, experiments where participants are asked to identify people they know just by looking at a moving silhouette have achieved a very high success rate. Starting out from a video of the subject walking, the newly-developed system distinguishes the background silhouette, which then becomes a sequence of silhouettes, placed one upon the other, resulting in a summary image. This final representation stores all physical characteristics and movements of each person walking, thus obtaining a unique imprint for every individual person. The drawback with the biometric technologies currently in use lies in the fact that both fingerprinting and facial recognition techniques require the subject to be close to a sensor and also to cooperate in the identification process, which s/he may not necessarily wish to do. The development of complementary techniques such as silhouette/gait analysis, which provides a way of getting round the need  to obtain an individual’s consent, is therefore crucial.

Coupling gait analysis with facial recognition

Ramon Mollineda warns however that, for the moment, the margin of error that gait recognition demonstrates in real scenarios outside a controlled environment means that this technique would be much more effective if combined with facial recognition. The Spanish academic points out that they are in fact “complementary methods: the way you walk can be detected from a distance without the need for a high-resolution image and can even be carried out against a backlight or with poor lighting; while face recognition has to be performed close-up using a high-resolution image.” Surveillance activity could therefore be carried out in a wider range of conditions. Moreover, if both methods can be applied in a given situation, the results – based on contrasting hypotheses generated by two separate biometric systems – are likely to be more reliable, underlines Ramon Mollineda.

By Pauline Trassard