基于双通道空洞卷积神经网络的手势识别  

Gesture recognition based on double-channel dilated CNN

作　　者：孙进[1,2] 张道周 张洋 习俊通 朱兴龙[1] SUN Jin;ZHANG Daozhou;ZHANG Yang;XI Juntong;ZHU Xinglong(College of Mechanical Engineering,Yangzhou University,Yangzhou 225127,China;State Key Laboratory of Mechanical System and Vibration,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]扬州大学机械工程学院,江苏扬州225127 [2]上海交通大学机械系统与振动国家重点实验室,上海200240

出　　处：《传感器与微系统》2022年第3期126-128,共3页Transducer and Microsystem Technologies

基　　金：国家自然科学基金面上资助项目(51475409);上海交通大学机械系统与振动国家重点实验室开放项目(MSV201810);江苏省研究生科研与实践创新计划项目(KYCX19_2102)。

摘　　要：为了提高手势识别过程中识别率,提出了一种基于双通道卷积神经网络(CNN)的识别算法。首先,对原始手势图像进行预处理,得到手部边缘图像;然后,分别选取手势图像和手部边缘图像作为CNN的两个输入通道;最后,在全连接层进行特征融合,并用SoftMax分类器对输出结果进行分类。通过实验证明:该算法能有效提高手势识别率,达到99.6%,增强了CNN的泛化能力。In order to improve the recognition rate in the process of gesture recognition,gesture recognition based on double-channels dilated convolution neural network(CNN)is proposed.Firstly,the original gesture image is preprocessed to get the hand edge image;then,the hand image and hand edge image are selected as the two input channels of CNN;finally,feature fusion is carried out in the full connection layer,and the output results are classified by SoftMax classifier.Experiments show that the algorithm can effectively improve the gesture recognition rate,reaching 99.6%,and enhance the generalization ability of CNN.

关 键 词：双通道 空洞卷积 神经网络 手势识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]