人臉辨識可以廣泛的應用在各個不同的領域之中，比方大樓門禁管制、犯罪人員辨識、金融提款的安全驗證，以及網路交易的身份識別等，都可以使用，不僅可靠，而且快速方便。 本論文使用結合PCA與臉部特徵的人臉辨識，其中包含使用PCA-based求出Eigenvalue、Eigenvector和Eigenspace，再藉由投影，得到輸入Sample與各訓練人臉的距離。此外，還藉由投影和邊緣偵測求得雙眼和嘴巴三個區域，以及雙眼、眼角和嘴巴七個特徵點，最後做正規化和特徵區域比對，然後與訓練之人臉做辨識。 結果驗證部分，本實驗使用Cohn Kanade Database，裡面包含了九十七位不同的人，每人各選十張，分成九組，其辨識率皆為100%。並且另外自行拍攝十個人，共六十張照片，來做訓練和辨識之用，其結果也皆可達到100%。 The Face recognition can be applied to many different fields. For example, the building entrance guard control, the criminal verification, the security verification of finance, and the identity verification of the network trade, etc. The face recognition is not only reliable but also fast and convenient. This disquisition combines PCA with face features for face recognition, including using PCA-based to figure out the Eigenvalue, Eigenvector and Eigenspace. Then get the distance between sample image and training image by the projection. Besides, we can get the eye and mouth areas, and the feature points of the eyes, canthi and the mouth by using the projection and edge detection. And finally, we do the normalization and area comparison to recognize the input image. The result verification. we use the Cohn Kanade Database that includes ninety-seven different persons. Ten pictures each person and we divide them into nine sets. All of the recognition rates are 100%. In addition, we make a face image database by ourselves. That includes ten different persons and sixty pictures to do the training and recognition. The recognition rate is also 100%
