基于边缘的双路卷积神经网络及其可视化  

An edge-based 2-channel convolutional neural network and its visualization

作　　者：李雨冲 闫昭帆 严国萍 LI Yu-chong;YAN Zhao-fan;YAN Guo-ping(School of Information Engineering,Chang’an University,Xi’an 710064,China)

机构地区：[1]长安大学信息工程学院

出　　处：《计算机工程与科学》2019年第10期1837-1845,共9页Computer Engineering & Science

基　　金：“弘毅长大”研究生科研创新实践项目(2018103,2018109)

摘　　要：为提高小尺度复杂图像识别准确率,通过对LeNet-5卷积神经网络并入一个新通道,让其处理与边缘有关的信息。结合两种通道产生的不同特征构造分类器,提出一种基于边缘的双路卷积神经网络,对小尺度复杂数据集进行识别。在包含10类产品数据上分类的结果表明,双路卷积神经网络的识别准确率远高于传统网络。最后通过神经网络可视化算法对双路卷积神经网络进行了可视化分析。In order to improve the recognition accuracy of small-scale complex images,an edge channel is added into LeNet-5 convolutional neural network to process the edge information.By combining the different features generated by two channels to construct a classifier,a 2-channel convolutional neural network is proposed to identify small-scale complex data sets.Classification results on ten types of product data show that the accuracy of the 2-channel convolutional neural network is much higher than that of the traditional network.Finally,the neural network visualization algorithm is adopted to visualize and analyze the 2-channel convolutional neural network.

关 键 词：图像模式识别 双路卷积神经网络 小尺度复杂图像 神经网络可视化 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]