作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2020, Vol. 46 ›› Issue (9): 268-273. doi: 10.19678/j.issn.1000-3428.0055648

• 图形图像处理 • 上一篇    下一篇

基于特征交换的CNN图像分类算法研究

生龙a,b, 马建飞a,b, 杨瑞欣a,b, 吴迪a,b   

  1. 河北工程大学 a. 信息与电气工程学院;b. 河北省安防信息感知与处理重点实验室, 河北 邯郸 056038
  • 收稿日期:2019-08-05 修回日期:2019-10-21 发布日期:2019-11-05
  • 作者简介:生龙(1982-),男,副教授、博士,主研方向为人工神经网络、图像处理、机器学习;马建飞、杨瑞欣,硕士;吴迪,副教授、博士。
  • 基金资助:
    国家自然科学基金"面向移动环境的情感推荐隐式反馈偏好挖掘研究"(61802107);河北省自然科学基金"公共环境下的WBANs共存技术研究"(1721203048);河北省教育厅基金"增量序列模式匹配下网络入侵检测方法研究"(ZD2018087);河北省教育厅基金"WBANs多网共存中MAC机制的融合与优化研究"(F2018402251)。

Research on CNN Image Classification Algorithm Based on Feature Exchange

SHENG Longa,b, MA Jianfeia,b, YANG Ruixina,b, WU Dia,b   

  1. a. School of Information and Electrical Engineering;b. Hebei Key Laboratory of Security and Protection Information Sensing and Processing, Hebei University of Engineering, Handan, Hebei 056038, China
  • Received:2019-08-05 Revised:2019-10-21 Published:2019-11-05

摘要: 针对深度学习在图像识别任务中过分依赖标注数据的问题,提出一种基于特征交换的卷积神经网络(CNN)图像分类算法。结合CNN的特征提取方式与全卷积神经网络的像素位置预测功能,将CNN卷积层提取出的特征图与同类标签特征图进行交换,充分融合有限的图像特征,以解决图像识别中样本不足的问题。实验结果表明,该算法对标注数据的依赖性较低且有效提升了网络识别准确率,适用于数据量较小的图像分类场景。

关键词: 深度学习, 卷积神经网络, 特征提取, 图像识别, 特征融合

Abstract: To address the problem that deep learning relies too much on labeled data in image recognition applications,this paper proposes a Convolutional Neural Network(CNN) image classification algorithm based on feature exchange.By combining the feature extraction method of CNN with the pixel position prediction function of full convolutional neural network,the feature map extracted from the convolution layer of CNN is exchanged with the similar label feature map,so the limited image features are fully fused to solve the lack of samples in image recognition.Experimental results show that the proposed algorithm can reduce the dependence on labeled data and significantly improve the recognition accuracy of network.It is appropriate for the image classification scenarios where data cannot be obtained in large quantities.

Key words: deep learning, Convolutional Neural Network(CNN), feature extraction, image recognition, feature fusion

中图分类号: