期刊名称：International Journal of Computer Science & Technology
印刷版ISSN：2229-4333
电子版ISSN：0976-8491
出版年度：2014
卷号：5
期号：1
语种：English
出版社：Ayushmaan Technologies
摘要：Due to its usefulness in recognition and identification, face recognition system proves its significance in real time applications. Such system proves to be useful in person’s authentication as a Biometric security system. In this paper we have proposed a face recognition technique for real time video. We proposed an improved PCA-LDA based face recognition algorithm with feature extraction using Haar Discrete Wavelet Transform and DCT. The 2-D Discrete Haar Wavelet transform and DCT is adopted to extract low frequency component of image. Then PCA method is used for reduction of redundant information which is due to directional vectors. To discriminate the class present in database, additionally LDA algorithm is used. Our proposed algorithm reduces the complexity due to reduction of redundant information and by use of LDA the discriminating performance is achieved. The result shows that face recognition is achieved for frontal face and also for directional faces (up to 250-300). We succeed for face recognition for low illumination effect. We have achieved an appreciable results of recognition 77.5% for DWT and 75% for DCT in real-time.
