Face recognition has become one of the important research areas and is used in a wide range of applications. In addition to accuracy, traditional face recognition methods face challenges on time-consuming to identify and apply to distributed systems in a large data environment. To solve these problems, we proposed a facial recognition method using the Gabor wavelet technique and the MapReduce parallel processing model. We performed parallel processing at the extraction and recognition stage with the MapReduce model in the Spark environment. Experimental results show that the proposed method significantly improves the computing time and the accuracy of face recognition.