Secure Authentication System Using Video Surveillance

Biometric person recognition for secure access to restricted data/services using PC with internet connection. To study, an application PC to be used as a biometric capturing device that captures the video and recognition can be performed during a standard web session. The main contribution of this novel proposal is, making comparison of portrait. Centroid context algorithm is used for selecting an random movements from an video and stored it in a database and make them to compare with video. To better characterize a portrait in a sequence, triangulate it into triangular meshes, which we extract the features: skeleton feature and centroid feature. Skeleton feature and centroid context feature working together makes human movement analysis a very efficient and accurate process. Depth first search(DFS) scheme is used to extract the skeletal feature of a portrait from triangulation result, from skeletal feature result, centroid context feature is extracted, which is a finer representation that can characterize the shape of a whole movements. For efficient and accurate process, generate a set of key portrait from a movement sequence. The ordered key portrait sequence is represented by string. For arbitrary matching action, string matching algorithm is used for implementing the concept.