通信工程文献翻译——由于有限的可用样本，人脸识别是一个典型的不适定问题。实验结果显示，结合多分类器融合的最类似的RBPCA方法，结合协方差矩阵正则化和基于快的主成分分析，能为人脸识别提供一个有效的框架，减轻小样本容量的问题。

随着生物统计学系统的出现，人脸识别在安全和监控应用中被广泛应用。在生物统计学中最具挑战性的任务之一就是人脸识别，包括在一个已知人脸库中将一个输入人脸与个别条目进行联系起来。人脸识别过程包括两个主要的步骤：特征提取和特征分类，包括整体的、基于特征的以及混合匹配方法。在人脸识别中的主要技术问题，如变化性大，复杂的非线性增加剧烈，高维数以及比较少的样本容量是所以常见的存在的问题。

我们提出了一个人脸识别系统，应用现阶段成熟的最类似的RBPCA技术，通过结合分类器来提高分类性能。它是一个统计的分类器，结合了协方差矩阵正则化、基于分块的主成分分析以及最大似然分类器。就如我们所知道的，这是研究者第一次将最类似的RBPCA和信息融合方法结合起来解决人脸识别的问题。

通过为人脸图像创建一个统计模型，来对人脸识别创建一个异常的分类器系统。通过最大似然法得到有监督参数的分类器和贝叶斯分类器是人脸识别中的主要任务。换而言之，为统计的人脸识别提供一个新的框架。在决策规则中，还将一个分类器结合方案应用于合并更多相关的信息，减轻小样本容量的问题。与其他广泛使用的人脸识别分类器方法相比较，......

Face recognition is typically an ill-posed problem because of the limited number of 

available samples. As experimental results show, combining multiclassifier fusion with the 

RBPCA MaxLike approach, which couples covariance matrix regularization and blockbased principal component analysis (BPCA), can provide an effective framework for face 

recognition that alleviates the small sample size problem.

With the advent of biometric systems, the human face has been widely exploited in security and surveillance applications. One of the most challenging tasks in biometrics is humanface recognition, which consists of linking an input face with an individual entry in a database of known persons. The face recognition process involves two main steps: feature extraction and feature classification, which includes holistic, feature-based, and hybrid matching methods.1The key technical problems in face recognition— large variability, highly complex nonlinear manifolds, high dimensionality, and small sample size2—are common to all existing approaches.

We propose a face recognition system that improves classification performance by combining classifier combination strategies with the recently developed RBPCA MaxLike technique,3 a statistical classifier that combines covariance matrix regularization, block-based principal component analysis (BPCA), and maximum-likelihood classification.4 To our knowledge, this is the first......