Silhouette Analysis-Based Gait Recognition for Human Identification
IEEE Transactions on Pattern Analysis and Machine Intelligence
Video-Based Human Motion Estimation System
ICDHM '09 Proceedings of the 2nd International Conference on Digital Human Modeling: Held as Part of HCI International 2009
| Hi-index | 0.00 |
The paper presents a system for human action recognition using contour based shape representation. With the rapid progress of computing and communication technology smart user computer interfaces are becoming most widespread. A major goal is to go further than traditional human computer interaction (like mouse or keyboard) and to find more natural means of interaction with computers, including the application of computer games and surveillance. The objective of this work is to achieve representation eigenspace for modeling and classifying actions performed by individuals. Eigenspace is the subspace for each type of action. A representation eigenspace approach based on the Principal Component Analysis (PCA) algorithm is used to train the classifier. Behaviors are classified with respect to a predefined set of learning actions. The key points of this approach include the mode silhouettes are extracted from video, the kind of shape descriptor used, the development of the new eigenspace and the kind of classification used. Performance of the system is expressed in terms of percentage of right or wrong classifications.