面向人脸识别的最大间距准则优化研究  

Optimization of maximum margin criterion for face recognition

作　　者：张仕霞 金聪[1] Zhang Shixia;Jin Cong(School of Computer, Central China Normal University, Wuhan 430079, Chin)

机构地区：[1]华中师范大学计算机学院

出　　处：《电子测量技术》2018年第8期128-134,共7页Electronic Measurement Technology

基　　金：国家社会科学基金(13BTQ050)项目资助

摘　　要：基于L1范数的二维最大间距准则算法直接求解十分复杂,即使采用贪心算法将原问题简化,求解方法的复杂度仍然很高,而且不能得到全局最优解。针对此问题提出了新的求解方法,将粒子群优化算法与基于L1范数的二维最大间距准则算法有效的结合,利用粒子群算法优化降维的投影矩阵,同时利用L1范数的二维最大间距准则构造粒子群的适应度函数,使算法能够得到全局最优解,提高人脸识别的精度。改进算法在ORL、Yale以及加噪数据集上的实验验证了该算法的有效性。The solution of the two-dimensional maximum spacing criterion algorithm based on L1 norm is very complicated.Even if the greedy algorithm is used to simplify the original problem,the complexity of the solution is still high and it can not be obtain the global optimal solution.In this paper,a new method is proposed to solve this problem.Since the particle swarm optimization algorithm can obtain the global optimal solution,this paper effectively combines the particle swarm optimization algorithm with the two-dimensional maximum spacing criterion algorithm based on L1 norm,The optimal algorithm is used to find the optimal projection matrix,which makes the algorithm get the global optimal solution and improve the efficiency of face recognition.The idea of this paper verifies the effectiveness of the algorithm in experiments with ORL and Yale and the noisy data set.

关 键 词：粒子群算法 二维最大间距准则 特征提取 人脸识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN911.73[自动化与计算机技术—计算机科学与技术]