Facial Recognition System in a Small Place with Speaker Validation

• Main Authors: Zhong-Ling Tsai, 蔡仲齡
• Other Authors: Jung-Hua Chou
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/91510232901807384588

碩士 === 國立成功大學 === 工程科學系碩博士班 === 96 === In this thesis, a facial recognition system with speaker validation in a small place is designed. This system is suitable for a small place that has people less than 20. There are three sub - systems in the system - the database system, facial image recognition system, and speaker validation system. First, use the database system to construct a facial image database and a speaker voice database which are the special database for the present system. Second, if the facial image recognition system judges that the user is a member of the database, the system allows that user to enter. Otherwise, the user has to be tested by the speaker voice validation system. The system lets the user pass, if the answer of the speaker voice validation system matches that of the facial image recognition system. In the system, the accuracy rate is 93% above. It costs about 0.5 seconds to recognize one person in the facial image recognition system, and 1 second in the speaker voice validation system.