基于高斯滤波器的人脸识别方法.docVIP

基于高斯滤波器的人脸识别方法 曲金帅[1]，赵明玺[1]，范菁[1]，何远斐[1] 云南省高校无线传感器网络重点实验室，云南 昆明，650031 摘要：人脸识别是根据某些标识对人进行身份识别，来达到监督、管理和识别目的的一种技术。近年来对这项技术进行了广泛而深入的研究，Gabor变换因其良好的时频局部化特性，能够提供最为实质的人脸特征、削弱噪声的干扰、减少计算量，因此将小波变换应用于人脸识别与检测具有良好的发展前景。由于人脸变化的诸多不确定因素，以及外部环境如光照等对成像系统的影响，使得人脸定位与识别具有极高的难度。另外人脸识别存在动态性和复杂性的特点，给数据处理带来诸多不便，限制了人脸识别的效率，本文借鉴了Gabor滤波和高斯滤波及人工神经网络的人脸识别的两种算法，并通过MATLAB编程实现了对其算法的验证，对实验数据计算得出了两个主要指标灵敏度和阳性预测值。结果前一种算法对人脸数较少的处理效果更好，后一种算法适合处理人脸数较多的图像。 关键词：人脸识别、Gabor滤波器、高斯滤波器 Face recognition method based on Gaussian filter Qu Jinshuai1，Chen Nan1，Fan Jing1，He Yuanfei[1] University Key Laboratory of Wireless Sensor Networks in Yunnan Province, Yunnan University of Nationalities，Kunming 650031, Yunnan, P.R.China ABSTRACT:Face recognition is based on the identification of some person identification, to achieve a technical oversight , management and identification purposes . In recent years, the technology has been extensive and in-depth research , Gabor transform its good time-frequency localization features to provide the most substantial facial features, and weaken the interference noise , reduce the computation , so the wavelet transform is applied Face recognition and detection has good prospects for development. Due to many uncertainties affecting the face changes , as well as the external environment , such as light and other imaging systems , making the face location and recognition with a high degree of difficulty. In addition there is a dynamic and complex recognition of the characteristics of data processing inconvenience to limit the recognition efficiency , the paper draws on two Gabor filtering and Gaussian filtering algorithms and artificial neural networks face recognition , MATLAB programming through its algorithm to validate the experimental data of the two main indicators calculated sensitivity and positive predictive value . Results Before an algorithm for face fewer treatment better, after a few more algorithm processing facial image fit. Keywords : face recognition , Gabor filt