Abstract: In order to take full advantage of multi-core processor resources, the parallel programming model by building blocks with multithreading was studied, hence improving the performance of the program. According to the integral structure of Principal Component Analysis (PCA)-based face recognition algorithm, a functional module named train() was designed for the training of recognizing generated samples in the environment of Matlab cluster. The parallelization of this algorithm was realized by task partition. The experimental results indicate that the stable recognition rate of 94.167% and the approximately linear speed-up ratio verify the correctness and high efficiency of the parallel algorithm.

Key words: face recognition, Principal Component Analysis (PCA), Matlab cluster, multi-core, task partition, parallel computing

摘要： 为了充分利用多核处理器资源,研究了多线程构建模块并行编程模式,从而提高程序的性能。在Matlab集群环境下对主成分分析(PCA)人脸识别算法设计了训练识别生成样本的功能模块train(),通过任务分割实现了算法的并行化。实验结果表明,94.167%的稳定识别率和趋近线性的加速比验证了并行算法的正确性和高效性。

关键词: 人脸识别, 主成分分析, Matlab集群, 多核, 任务分割, 并行计算