摘要
提出了采用低通过率波、去最小亮度和向量柱状图来提取人脸特征的方法,设计了模糊ART神经网络的结构、学习规则和识别算法,并采用模糊ART神经网络对向量柱状图生成的特征向量进行识别。仿真实验证明,通过调整神经网络的警戒参数值,不同的人具有不同的最大在线识别率,所有人平均的在线最大识别率可以达到89%。
The way of using low pass filter, minimum intensity subtraction and vector histogram to extract image feature of face were presented. The structure, learning rule and recognition algorithm of fuzzy ART neural network were described and designed. The fuzzy ART neural network was used to recognise the eigenvector generated by the vector histogram. Simulation experiment shows that there was different maximum online recognition rate＇for different person when the neural network vigilance parameter values are selected properly. The maximum average online recognition rate could be up to 89%.
出处
《计算机应用与软件》
CSCD
2010年第2期261-263,共3页
Computer Applications and Software
关键词
图像识别
神经网络
模糊ART
向量柱状图
识别算法
Image recognition Neural network Fuzzy ART Vector histogram Recognizing algorithm
