基于线性降维技术和BP神经网络的热红外人脸图像识别

结合线性降维技术与BP神经网络对热红外人脸图像进行了识别研究.首先利用主成分分析和线性判别分析对热红外人脸图像进行图像降维及特征提取,然后将提取出的热红外人脸图像特征向量用于BP神经网络的训练,得到一个鲁棒性和容错性较强的分类器,用这个分类器对热红外人脸图像进行分类识别.实验结果表明,由于所提方法在提取便于分类的模式特征基础上,采用神经网络作为分类器代替特征向量间的欧氏距离判别,获得了较高的热红外人脸图像识别率.

Thermal infrared face image recognition based on combination of linear dimensionality reduction technology and BP neural network

Thermal infrared face image recognition is studied by combining linear dimensionality reduction technology and BP neural network.First,the methods of principal component analysis and linear discriminant analysis are adopted to reduce the dimensionality of thermal infrared face images and extract feature vectors from thermal infrared face images.Then,the feature vectors are used for BP neural network training,and a classifier with strong robust and fault tolerance can be generated for thermal infrared face i...