摘要人脸识别是基于人的面部特征信息进行身份识别的一种生物识别技术,也是模式识 别的重要研究方向。不同人的面部特征普遍具有较大差异,并且人脸信息的采集过程简单 方便,所以人脸识别可以被用为验证身份的可靠依据,具有广泛的应用前景和重要的科研 价值。74268
稀疏表示是一种来源于压缩感知理论的信号表示方法,应用于人脸识别领域后因其 具有较高的鲁棒性收到广泛关注。本文采用稀疏表示分类方法进行人脸识别,并设计出图 形化界面,在不同的测试人脸数据库上进行识别实验。用于测试的图像库中包含了不同遮 挡、光照以及装饰等条件下的人脸照片,将这些测试图像进行降维、转换为灰度图像等处 理后,使它们由训练图像的线性组合来表示,在获得的组合系数中非零系数所在的类别即 为正确的分类。
毕业论文关键词 稀疏表示 人脸识别 图像预处理 特征提取
毕 业 设 计 说 明 书 外 文 摘 要
Title Face recognition based on sparse representation
Abstract
Face recognition is a biometric technology based on the identity of an inpidual’s facial features。 Due to the fact that facial features are usually easy to get and different facial characteristics are always quite different, face recognition can be used to identify the inpidual reliably and easily。 Because of its convenience and effectiveness, face recognition has a wonderful application prospect and important scientific research value。
Sparse representation is one kind of signal representation that originates in compression perception theory, which has received widespread attention because of its higher robustness since applied in the field of face recognition。 In this paper, the robust sparse representation classification method is used for face recognition, and a graphical interface is designed for identifying faces from different face databases。 Test images of faces are under varying occlusion, lighting and decoration conditions, firstly they should be down sampled and transferred to grayscale image, and then they are represented by linear combination of training images, the nonzero coefficients category is the correct classification。
Keywords Spare representation Face recognition Image preprocessing Feature extraction
本科毕业设计说明书 第 I 页
目 次
1 引言 1
1。1 研究背景及意义 1
1。2 人脸识别简介 1
1。3 现有人脸识别方法 1
1。4 本文主要内容和安排 2
2 稀疏表示分类算法 3
2。1 稀疏表示基本概念 3
2。2 L0 范数最小化问题 4
2。3 稀疏表示的特点 5
3 基于稀疏表示的人脸识别6
3。1 字典的构造 6
3。2 基于稀疏表示的人脸识别算法 7
4 基于 MATLAB 的人脸识别实现 9
4。1 实验 9