基于不变矩特征和神经网络的人脸识别模型

Face Recognition Based on Moment Invariants and Neural Networks

不变矩是图像的一种统计特征,具有平移不变性、旋转不变性和比例不变性,广泛应用于图像识别。该文将图像矩阵的不变矩作为识别特征,建立了人脸识别模型。将人脸图像经过不变矩特征提取、不变矩矢量标准化及不变矩矢量排列处理后,运用BP网络进行识别,经过竞争选择,获得识别结果。利用ORL人脸数据库进行仿真实验,结果表明,该文的人脸识别模型实现简单、识别率高、训练速度与识别速度较快、便于实时实现。

Moment invariant is invariable under shifting,scaling and rotation,and is widely applied in pattern recogni-tion.In this paper,a face recognition method is presented based on moment invariant ,in which moment invariant of image matrix is used as feature and BP net as recognition method.Face recognition model is established,which includes moment invariant feature extraction,moment invariant vector standardization,moment invariant vector arrangement ,BP net,and competition selection.Experiment results based on ORL face database demonstrate that the model designed has the characteristic of simple realization,high recognition rate,rapid training speed and recognition speed,and convenience for real-time realiza tion.Greater train ing errors are beneficial to improve training speed and efficiency of BP net,and profit real-time realization.