| Summary: | 碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 98 === In recent years, with progress of science and technology, personal identity and data protection becomes more important. Human face recognition that is one of the biometric identification methods has been widely used in the national safety and the government organizations. In this thesis, we use the compressed sensing theory to develop a human face recognition system. First, the face recognition problem is transformed to L1 optimum problem. Then, two methods are applied to solve the L1 optimum problem. One is called the log barrier method which can get the optimal solution but takes longer time; the other is the iterative hard threshold(IHT) method which is an iterative method to get the approximate solution. Next, combined with four different features of face, several recognition results are gotten. The face database used in this thesis is obtained from different lighting conditions. As the results, the recognition rate of the log barrier method is better than the IHT method, but it takes longer recognition time.
|
|---|
