中文摘要：

分析了KNN分类算法的流程,然后在K值的动态获取和分类加权两个方面对分类算法进行改进;利用MapReduce编程思想完成KNN分类算法在Hadoop集群环境下的移植和实现。实验数据证明,改进后的KNN分类算法在人脸识别精度、识别效率和稳定性3个方面得到了有效提高。

英文摘要：

KNN classification algorithm was analyzed with respect to its processes and then improved in dynamical acquisition and classified weighing of K value. MapReduce programming i- deas were adopted to complete the transplant and implementation of KNN classification algorithm in the Hadoop cluster environment. KNN classification algorithm experimental data demonstrate the improved face recognition accuracy, efficiency and stability.