摘要: |
针对核主元分析在人脸识别中的时间开销过大的不足,提出了一种核主元分析的DSP
实现方法。介绍了核主元分析算法原理,给出了DSP平台的硬件框架和程序优化方案。在DSP
硬件平台上的测试结果表明:系统运行稳定,与通用计算机平台相比,明显提高了识别速度,
同时保证了人脸识别率,具有一定的实用价值。 |
关键词: 人脸识别 特征提取 核主元分析 优化程序 数字信号处理 |
DOI: |
|
基金项目: |
|
A face recognition method based on KPCA and its DSP implementation |
LIU Song,TAN Jian-jun |
(College of Information Engineering,Hubei University for Nationalities,Enshi 445000,China) |
Abstract: |
In order to reduce recognition time of kernel principle component anal
ysis(KPCA) algorithm,the realization scheme of KPCA based on DSP is presented i
n this paper. The KPCA algorithm is introduced, hardware platform frame based on DSP a
nd program optimization scheme are given. The experimental results in DSP platfo
rm show that the system is of good stability,high recognition rate and high spee
d,practical value,and it shows obvious advantage over that of universal comput
er. |
Key words: face recognition feature extract kernel principle component analysis optimizatio
n program DSP |