基于ARM的人脸识别系统的研究与实现

Research and implementation of face recognition system based on ARM

使用人脸类Haar特征进行人脸检测,采用Daubechies小波去噪变换和PCA降维算法提取人脸特征,将经过PC机变换后的训练样本特征子空间文件通过网络传输到嵌入式平台,并结合最近邻算法识别人脸.实现了一种嵌入式人脸识别系统,解决了嵌入式人脸识别系统由于图像处理数据巨大而造成处理效率低的难点.基于MagicARM2410开发板实现了该系统,结合实际图片进行了人脸识别测试,实践结果表明系统效果良好.

The embedded face recognition system detects face by using Haar-like features of face, and the face features are extracted by using Daubechies wavelet filter and PCA( principal component analysis) dimension decline algorithm. The face features subspace of training samples is first acquised at personal computer,and then it is transmitted to the platform of embedded system by network,at last the test face is recognized by using the nearest distance algorithm. The system solves the problem that the efficiency of embedded face recognition system is low because of huge data in the image processing. The system is implemented based on MagicARM2410 development board,and the performance of face recognition was tested by using actual images. The results indicate that the recognition performance of this system is good.