基于神经网络的掌纹识别技术研究

本文建立一种基于带有Sigmoid激活函数的神经网络的掌纹识别方法。基于神经网络对样本训练数量的要求,对图像库进行提升样本数量的预处理,增加样本容量;然后,采用多方向卷积核组建的神经网络,提取掌纹特征;最后,利用非线性函数输出结构优化输出。实验结果表明所提出方法降低了时间复杂度的同时,提高了掌纹识别率。

A palmprint recognition method based on neural network with sigmoid activation function is proposed in this paper.Since the training number of samples is required by neural networks,the preprocessing of palmprint images is required to increase the sample size;The neural network is constructed by using a multiple direction convolution kernel to extract palmprint features.Finally,the output is optimized by nonlinear function output.Experimental results show that the proposed method reduces the time complexity and improves the palmprint recognition rate.