A Hybrid Face Recognition System for Managing Time of Going to Work and Getting away from Office

Abstract

A handy method of individual identification based on the face recognition was examined aiming at the construction of time management system of going to work and getting away from office. The foreground was easily extracted by the background difference method to acquire the facial image in the fixed place, and the face region was able to be specified by a flesh-colored extraction with the filter. Individual identification was tried by specifying the positions of the face parts, generating the characteristic vector of ten dimensions, and using fuzzy reasoning from the position in the principal ingredient score space. As a result, it turned out that a difficult group of individual identification existed. For such a group, it was confirmed that the support vector machine was effective to identification. Then, an hybrid identification method was proposed.