Face Recognition Using SIFT and PCA

• Main Authors: Tzu-Yen Shu, 許子彥
• Other Authors: Cheng-Yuan Tang
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/36246708718401130979

碩士 === 華梵大學 === 資訊管理學系碩士班 === 96 === There are lots of methods proposed for face recognition in the literature. The images capturing faces in different viewpoints (frontal face and the side view of face) may rapidly decrease the accuracy of face recognition. It is a well-known difficult problem. In this paper, we try to propose methods to increase the accuracy rate. In this paper, the Principal Component Analysis (PCA) is used to get the face candidates in a face database. It is well-known that the Scale Invariant Feature Transform (SIFT) for matching is invariant to scale and rotation. In this paper, SIFT is used for recognition from the face candidates extracted from PCA. There are still some match errors in the results after processing SIFT in our experiment. Therefore, two matching methods using the y-axis distance and cross product for improving the recognition rate are proposed in this paper. There are 12 persons, each having 9 images in different viewpoints in our face database. In our face database, Using SIFT and PCA, the recognition rate is 95.8%. By adding the proposed matching methods of the y-axis distance and the cross product, the recognition rate is up to 98.6%. In the future, we will test our proposed methods to face databases having more persons.