一种模块化神经网络的人耳识别方法

提出了一种基于模块化神经网络的人耳识别方法。对人耳图像进行一系列的预处理后,采用PCA方法对图像进行特征提取。构建了模块化神经网络模型,并用分层遗传算法对该模型进行优化,选择训练阶段样本和测试阶段样本对人耳图像进行训练和测试,得出识别率。实验结果表明,基于模块化神经网络的人耳识别相对于传统的神经网络优化了设计参数,得到最优体系结构,提高了人耳识别率。

Ear recognition method based on modular neural networks

A new ear recognition method based on module neural network was presented in this paper. By us-ing PCA method,the characteristic of the human ear image was extracted after a series of pretreatment on the ear image. The modular neural network model was built up and optimized by using hierarchical genetic algo-rithm. Training samples and testing samples were selected to train and test the human ear images,finally came up with recognition rate. The experimental results show that the ear recognition based on modular neural net-work optimizes the design parameters compared with the traditional neural network. The optimal system struc-ture is obtained,and the human ear recognition rate accordingly improved.