基于GPU并行加速的2DPCA人脸识别算法
首发时间:2014-12-19
摘要:人脸识别技术是一种基于人的脸部特征信息进行身份识别的一种识别技术。如今,人脸识别技术已经成为当今社会不可或缺的一个安全技术。主成分分析(PCA)算法是人脸识别的主要算法之一,图片的像素较高并且训练集中的图片较多导致矩阵计算量大,因此基于PCA算法的人脸识别耗时较长,不适合单机串行运行。本文针对PCA算法的这一缺点,综合了基于二维矩阵处理的二维主成分分析(2DPCA)算法,提出了一种单机基于GPU并行的改进2DPCA算法。通过ORL人脸库进行实验证明,改进后的算法能够在仅仅使用单机的前提下,将单线程的PCA算法速率提高了超过10000倍,并且识别率保持相当。
关键词: 人脸识别 主成分分析 二维主成分分析 GPU并行?????
For information in English, please click here
A GPU Based Speedup 2DPCA Face Recognition Algorithm
Abstract:Face recognition technology is based on identifying by using human face features. Nowadays, it has become an indispensible part of our society. Principle Component Analysis (PCA) algorithm is one of the major face recognition technologies. With thousands of human faces to recognize and pixels to read, the matrix computing becomes huge, which makes PCA algorithm cost long time and undoubtedly unfit for serial running. To overcome the shortcoming, the paper combines a Two-Dimensional PCA (2DPCA) algorithm, which based on two-dimensional processing, and discovers a new parallel speedup 2DPCA algorithm using GPU computing which can still run on a single computer. Through the experiments by testing ORL human faces, a conclusion is brought out that the new algorithm is more than 10,000 times faster than the original PCA algorithm, while still using one single computer and can keep the recognition rate.
Keywords: Face recognition Principle Component Analysis Two-Dimensional Principle Component Analysis GPU parallel computing.
基金:
论文图表:
引用
No.4621833102339214****
同行评议
勘误表
基于GPU并行加速的2DPCA人脸识别算法
评论
全部评论0/1000