摘要
基于压缩感知的人脸识别算法（SRC）利用了高维数据分布的稀疏性进行建模，能够很好地解决图像高维处理问题，有效地避免维数灾难。对基于压缩感知的人脸识别算法的基本原理进行了深入地分析和研究，并对SRC算法进行了改进，提出了基于分块思想的压缩感知人脸识别算法，解决了人脸图像识别中存在的遮挡问题，避免了特征提取过程所造成的图像信息丢失，也避免了图像中局部信息的损坏对整体识别效果的影响。通过仿真实验表明改进算法的识别率比SRC算法的识别率提高了7％～8％。
The face recognition algorithm based on compressed sensing （SRC） uses the sparsity of high-dimensional data distribu- tion to perform modeling, which can solve the problem of high-dimensional image processing and effectively avoid dimension disaster. This paper analyzes the basic principle of face recognition algorithm based on compressed sensing in depth,improves the SRC algorithm and puts forward the compressed sensing face recognition algorithm based on blocking idea. It solves the occlusion problem in the face recognition, avoids the loss of image information in the process of feature extraction, and a/so avoids the impact of local information dam- age on overall recognition effect. Simulation experiments show that the recognition rate of improved algorithm is 7% - 8% higher than SRC algorithm.
出处
《无线电通信技术》
2013年第5期93-96,共4页
Radio Communications Technology
关键词
人脸识别
压缩感知
分块
遮挡
face recognition
compressed sensing
blocking
occlusion
