AbstractWith the significant increasing and development of computers, huge use of high-tech mobile devices, vision-based facial recognition has improved significantly. However, the performance of computers is still lower than human performance, since humans have a more prominent ability in terms of challenging environments such as occlusion or variations. Inspired by humans’ recognition method, which includes both holistic and local features, we propose a dual-stage facial recognition method that utilized both holistic and local features-based recognition algorithms. The first stage Principal Components Analysis (PCA) is utilized to recognize the test image coarsely. If the confidence level test passes, the recognition process will be terminated. Otherwise, the second stage where High Dimensional Local Binary Patterns (HDLBP) are employed will be pursued. The contribution of our work is proposed a hybrid algorithm with flexible structure that ensures fast and accurate results. An experiment of this dual-stage method is performed with a CMU-PIE database to identify the effectiveness and validity of the research, and we obtain better recognition results under a variety of illuminations not only in terms of the computation speed but also in terms of recognition rate in contrast to PCA and HDLBP based recognition algorithms.Keywords: Dual-stage, PCA, High-Dimensional-LBP, Face Recognition