小波神经网络在人脸识别中的应用

摘要

Face recognition is a front complex subject, which includes physiology, psychology, image processing, computer vision, pattern recognition and mathematics,etc. As a research success in the are-a of wavelet analysis theory, Wavelet Neural Network, a feed-forward network, based on the foundation of reliable theory, avoids the blindness in structure design of BP neural network, excludes the probability of sub-optimization in local non-linear optimization problems during network training process and has the capabilities of function learning and generalization. This paper presents a face recognition algorithm based on wavelet neural network, which depends on the multi-resolution property of wavelet and the robustness and memorization features of neural network, combined with the wavelet neural network step adjustment algorithm, a wavelet neural network is designed for the use of face recognition, whose effectiveness and accuracy are verified by some experiments.%人脸识别是一个涉及生理学、心理学、图像处理、计算机视觉、模式识别和数学等多个学科的前沿课题.小波神经网络是在小波分析研究获得突破的基础上提出的一种前馈性网络,避免了BP网络等结构设计上的盲目性,网络训练过程从根本上避免了局部最优等非线性优化问题,有较强的函数学习能力和推广能力.基于小波神经网络,文中提出了一种新的人脸识别算法.该算法利用小波多分辨特性和神经网络的鲁棒性和记忆性,同时结合了加速网络收敛速度的小波神经网络步长调整算法.实验证明该算法有高的检测率和有效性.