基于BP网络的人脸朝向识别模型  

Face Orientation Recognition Based on BP Neural Network

作　　者：吴育文[1] 陈琛[2] 康文豪[3] 

机构地区：[1]东北大学秦皇岛分校信息与计算科学系,秦皇岛066004 [2]大连交通大学软件学院,大连116001 [3]浙江科技学院信息与计算科学系,杭州310013

出　　处：《影像技术》2012年第1期29-32,33,34,64,共7页Image Technology

摘　　要：本文基于BP神经网络建立数学模型解决人脸朝向识别这一重要问题。通过对图片压缩、标准化处理等步骤,建立了一个全新的BP神经网络。并用MATLAB对50张图片进行了实验,实验结果表明,该方法识别准确率高、识别过程速度快。最后,我们就训练样本数目的确定与更有效识别方法的探索两个方面对模型提出了改进。We build a Mathematical model to solve the important problem of face orientation recognition based on BP neural network.By compression and standardization of images, a new Bp neural network is established. The recognition using Matlab software and 50 pieces of picture is simulated. The experiments show that this system has a high rate of accuracy, quick speed of recognition. At last,We have made improvements of the model in terms of certainty of the amount of training samples and exploring better identification method.

关 键 词：BP神经网络 人脸朝向识别 主成分分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]