| Summary: | 碩士 === 國立高雄應用科技大學 === 資訊工程系 === 100 === With the rapid increase of the member of Facebook [1], Twitter [2], the rapid development in the social network, as well as the popularity of digital cameras, upload photos to be managed on the social network, has become an important, in order to upload photos to the social network become more efficient management, how to improve the automatic identification of people face annotation, has become popular topics in recent years.
This thesis applies face recognition to identification the face in the photo on uploaded to the social network , before facial recognition applications on the social network to do identification, the first step is normalization, this paper proposed method is not required, just do the face recognition after the face detected.
In the recognition process, the application of concepts to friends as a classifier on the social network, the use of the LDA to build a discriminative feature space when upload new photos through each classifier to make a determine, then combine the final determine results to enhance the accuracy of face recognition.
In this thesis, use two database as the experimental test, facebook database is download from facebook album. In order to compare with another research, we use LFW database to most of the face recognition research will be used and convenience compared with other research results.
|
|---|
