eduzhai > Applied Sciences > Engineering >

Person Re-identification by analyzing Dynamic Variations in Gait Sequences

  • Save

... pages left unread,continue reading

Document pages: 12 pages

Abstract: Gait recognition is a biometric technology that identifies individuals in avideo sequence by analysing their style of walking or limb movement. However,this identification is generally sensitive to appearance changes andconventional feature descriptors such as Gait Energy Image (GEI) lose some ofthe dynamic information in the gait sequence. Active Energy Image (AEI) focusesmore on dynamic motion changes than GEI and is more suited to deal withappearance changes. We propose a new approach, which allows recognizing peopleby analysing the dynamic motion variations and identifying people without usinga database of predicted changes. In the proposed method, the active energyimage is calculated by averaging the difference frames of the silhouettesequence and divided into multiple segments. Affine moment invariants arecomputed as gait features for each section. Next, matching weights arecalculated based on the similarity between extracted features and those in thedatabase. Finally, the subject is identified by the weighted combination ofsimilarities in all segments. The CASIA-B Gait Database is used as theprincipal dataset for the experimental analysis.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...