Abstract | ||
---|---|---|
Gait recognition is an unobtrusive biometric, which allows identification of people from a distance by the manner in which they walk. In this paper, a new approach is proposed for extracting human gait features based on body joint identification from human silhouette images. In the proposed approach, the human silhouette image is first enhanced to remove the artifacts before it is divided into eight segments according to a priori knowledge of human body proportion. Next, the body joints which act as the pivot points in human gait are automatically identified and the joint trajectories are computed. To assess the performance of the extracted gait features, fuzzy k-nearest neighbor classification technique is used to identify subjects from the SOTON covariate database. The experimental results have shown that the gait features extracted using the proposed approach are effective as the recognition rate has been improved. |
Year | DOI | Venue |
---|---|---|
2011 | 10.1007/978-3-642-25191-7_24 | IVIC (1) |
Keywords | Field | DocType |
body joint identification,human gait,joint trajectory,automatic body,improved gait recognition,new approach,body joint,gait recognition,gait feature,human body proportion,human silhouette image | Computer vision,Covariate,Pattern recognition,Gait,Computer science,Silhouette,Fuzzy logic,A priori and a posteriori,Artificial intelligence,Gait (human),Biometrics,Body joints | Conference |
Volume | ISSN | Citations |
7066 | 0302-9743 | 1 |
PageRank | References | Authors |
0.35 | 12 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Tze-Wei Yeoh | 1 | 6 | 1.43 |
Wooi-Haw Tan | 2 | 32 | 6.38 |
Hu Ng | 3 | 9 | 2.86 |
Hau-Lee Tong | 4 | 21 | 5.15 |
Chee-Pun Ooi | 5 | 2 | 2.05 |