Performing Content-Based Retrieval of Humans Using Gait Biometrics

Abstract

Abstract. In order to analyse surveillance video, we need to efficiently explore large datasets containing videos of walking humans. At survei llance-image resolution, the human walk (their gait) can be determined automatically, and more readily than other features such as the face. Effective analysis of such data relies on retrieval of video data which has been enriched using semantic annotations. A manual annotation process is time-consuming and prone to error due to subject bias. We explore the content-based retrieval of videos containing walking subjects, using semantic queries. We evaluate current biometric research using gait, unique in its effectiveness at recognising people at a distance. We introduce a set of semantic traits discernible by humans at a distance, outlining their psychological validity. Working under the premise that similarity of the chosen gait signature implies similarity of certain semantic traits we perform a set of semantic retrieval experiments using popular latent semantic analysis techniques from the information retrieval community.

Identifying humans using comparative descriptions

“…Samangooei and Nixon [1] developed a soft biometric system which used 23 absolute categorical labels to describe a subject's physical appearance. The 23 traits were chosen to be universal, distinct, easily discernible at a distance and largely permanent.…”

“…The gait database includes videos of 100 people walking in a plane normal to the view of the camera. Previously absolute categorical labels had been collected for the same database [1] -allowing comparisons between the two forms of description. The videos were displayed within a website shown on a computer monitor.…”

“…Initial analysis directly compared the comparative annotations with the absolute categorical labels gathered by Samangooei and Nixon [1]. This comparison between annotation techniques will not show which is better, only how much each technique differs from the other.…”

“…Although each attribute can have reduced discriminative capability, they can be combined for identification [1,2] and fusion with traditional 'hard' biometrics [3,4]. One of the main advantages of soft biometrics is their relationship with human description; humans naturally use soft biometric traits to identify and describe one an other.…”

“…However, a detailed Figure 1. Surveillance frame displaying common surveillance problems 1 human description of the suspects can still be determined especially when viewing the video from which this frame was derived. Soft biometric traits can be obtained from the data derived from low quality sensors, including surveillance cameras.…”