基于树型稀疏模型的重构算法在人脸图像中的研究  

RESEARCH ON RECONSTRUCTION ALGORITHM FOR FACE IMAGES BASED ON TREE SPARSE MODEL

作　　者：孙翠华[1] 

机构地区：[1]苏州农业职业技术学院,江苏苏州215008

出　　处：《计算机应用与软件》2015年第8期324-327,共4页Computer Applications and Software

摘　　要：人脸图像的识别率一直以来都是研究的热点。提出一种基于树状稀疏模型的算法。首先通过在小波域稀疏特性基础上构建一种图像数据的树状稀疏模型,进而针对此模型设计一种新的人脸图像重构算法。在人脸图像中使用消除人脸部的冗余信息,通过与传统重构算法的比较以及采用3个标准人脸数据集对该算法进行测试。仿真实验表明,该算法相对于参比算法,提高了人脸图像识别精度,具有很好的鲁棒性。The recognition rate of face images has always been the hot topic of research. In this paper,we present a tree sparse modelbased algorithm. First,we build a tree sparse model of image data based on the wavelet domain sparse features,and then we design a new face image reconstruction algorithm for this model. By applying it in face images the redundant information of the faces is eliminated. The algorithm is tested by comparing it with traditional reconstruction algorithm and adopting three standard face data sets. Simulation experiment shows that the proposed algorithm,compared with the reference algorithm,improves the precision of face image recognition and has very good robustness.

关 键 词：树状稀疏模型 重构算法 人脸图像识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]