摘要
针对人脸图像数据量大和BP神经网络容易陷入局部最优的问题,提出将主分量分析、遗传算法和BP神经网络相结合的新算法,并将其应用于人脸识别中。利用主分量分析法来处理人脸图像,以减小人脸图像的数据量,再利用遗传算法优化BP网络的初始权值和阈值,最后利用ORL数据库对该算法进行验证。实验结果表明,该算法可以大大减少人脸图像的数据量使收敛速度加快,并且可以克服其陷入局部最优的缺点,提高了识别精度。
In this paper, a new algorithm was applied into face recognition, which combined the principal component analysis, genetic algorithm and BP neural network. Principal component analysis was used to manage face images to reduce the amount of data, and genetic algorithm was used to optimize the initial weights and thresholds of BP network. Finally ORL database was used to verify the algorithm. Experimental results showed that it can greatly reduce the amount of data, speed up the convergence rate, and can improve the precision of recognition.
出处
《中国西部科技》
2015年第6期73-75,15,共4页
Science and Technology of West China
关键词
人脸识别
主分量分析
遗传算法
BP神经网络
Face Recognition
Principal Component Analysis
Genetic Algorithm
BP Neural Network