Parallel Facial Recognition System Based on 2DHMM

The constantly growing amount of digital data more and more often requires applying increasingly efficient systems for processing them. Increase the performance of individual processors has reached its upper limit therefore we need to build multiprocessor systems. To exploit the potential of such systems, it is necessary to use parallel computing, i.e. creating computer systems based on parallel programming. In practice, most often their used adjusts the parallelization of the data processes, or regulates the parallelization of the query tasks. The system of face recognition that requires high computational power is one of potential application of the computations parallelization, especially for large database sizes. The aim of the research was to develop a parallel system of face recognition based on two-dimensional hidden Markov models. The results show that compared to sequential calculations, the best results were obtained for parallelization of tasks, and acceleration for training mode was 3.3 and for test mode - 2.8.