Holistic-based methods, especially principal component analysis(PCA) and linear discriminant analysis(LDA), are very popularly used in face recognition. In last several years, the direct LDA(D-LDA) method was suggested to overcome the so-call “small simple size” problem. Besides, the researchers suggest bidirectional PCA(BDPCA) method, which may be more real-time and effective than PCA, and proposed BDPCA plus LDA (BDPCA+LDA). In the same way, we suggest bidirectional D-LDA(BDDLDA) should be faster than D-LDA, and proposed BDPCA plus BDDLDA (BDPCA+BDDLDA). In this paper, we present a real-time face recognition system for operating system authentication based on BDPCA+BDDLDA. A face database, which was collected with a CCD camera on the laptop, is used to evaluate this system. Experimental results show that this practical system is real-time and effective.