首页 > 论文 > 激光与光电子学进展 > 56卷 > 16期(pp:160101--1)

基于多尺度卷积特征融合的台风等级分类模型

Typhoon Classification Model Based on Multi-Scale Convolution Feature Fusion

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

为提高卷积神经网络对图像多尺度变化的感知能力,增加网络的尺度不变性,提出一种基于多尺度卷积特征融合的台风等级分类模型。在卷积神经网络中添加多尺度感知层,对卷积特征进行多尺度感知并进行级联。将多尺度正则化项添加到损失函数中,通过反向传播来最小化隐含层权重的残差,优化模型的特征提取能力。最后将多尺度高层语义特征通过Softmax分类层归一化成各图像类别的概率值,取最大概率值为最后图像的分类结果。为有效验证本模型的多尺度感知能力,选用红外卫星台风云图作为数据集,实验结果表明,本模型能有效感知并提取台风云图的局部特征,并在两个通用数据集MNIST和CIFAR-10上验证了本模型的泛化能力。

Abstract

In order to enhance the perception for the multi-scale image variation and improve the scale invariance of convolutional neural networks,this study proposes a typhoon classification model based on multi-scale convolutional feature fusion. A multi-scale perception layer is added to convolutional neural networks; then, convolutional features are multi-scale perceived and cascaded. A multi-scale regularization term is then incorporated into the loss function. The residual error of hidden layer weight is minimized and the feature extraction ability is optimized with backpropagation. Finally, multi-scale high-level semantic features are normalized to the probability value of each category using Softmax. The maximum probability value is used as the final classification result of the image. Infrared satellite cloud images are used as the dataset in our experiments to validate the multi-scale perception ability of the model. Experimental results show that the model can effectively perceive and extract the local features of the typhoon cloud map. The generalization ability of the model is verified using two general datasets, i.e., MNIST and CIFAR-10.

Newport宣传-MKS新实验室计划
补充资料

DOI:10.3788/LOP56.160101

所属栏目:大气光学与海洋光学

基金项目:国家自然科学基金(41501419);

收稿日期:2019-01-28

修改稿日期:2019-03-21

网络出版日期:2019-08-01

作者单位    点击查看

卢鹏:上海海洋大学信息学院, 上海 201306
邹佩岐:上海海洋大学信息学院, 上海 201306
邹国良:上海海洋大学信息学院, 上海 201306

联系人作者:卢鹏, 邹国良(plu@shou.edu.cn, glzou@shou.edu.cn)

备注:国家自然科学基金(41501419);

【1】Huang S P, Yu L and Feng Z Y. A survey on image categorization using supervector coding Electronic Technology & Software Engineering. 2013(4), 32-37(0).
黄双萍, 俞龙, 冯子勇. 基于超级向量编码的图像分类系统研究 电子技术与软件工程. 2013(4), 32-37(0).

【2】Krizhevsky A, Sutskever I and Hinton G E. ImageNet classification with deep convolutional neural networks. Communications of the ACM. 60(6), 84-90(2017).

【3】Huang H, He K, Zheng X L et al. Spatial-spectral feature extraction of hyperspectral image based on deep learning. Laser & Optoelectronics Progress. 54(10), (2017).
黄鸿, 何凯, 郑新磊 等. 基于深度学习的高光谱图像空-谱联合特征提取. 激光与光电子学进展. 54(10), (2017).

【4】LeCun Y, Bottou L, Bengio Y et al. . Gradient-based learning applied to document recognition. Proceedings of the IEEE. 86(11), 2278-2324(1998).

【5】Zeiler M D and Fergus R. Visualizing and understanding convolutional networks. ∥ Fleet D, Pajdla T, Schiele B, et al. Computer vision-ECCV 2014. Lecture notes in computer science. Cham:Springer. 8689, 818-833(2014).

【6】Luo J H and Wu J X. A survey on fine-grained image categorization using deep convolutional features. Acta Automatica Sinica. 43(8), 1306-1318(2017).
罗建豪, 吴建鑫. 基于深度卷积特征的细粒度图像分类研究综述. 自动化学报. 43(8), 1306-1318(2017).

【7】Wang L L, Liu J H and Fu X M. Facial expression recognition based on fusion of local features and deep belief network. Laser & Optoelectronics Progress. 55(1), (2018).
王琳琳, 刘敬浩, 付晓梅. 融合局部特征与深度置信网络的人脸表情识别. 激光与光电子学进展. 55(1), (2018).

【8】Niu L Q, Chen X Z, Zhang S N et al. Model construction and performance analysis for deep consecutive convolutional neural network. Journal of Shenyang University of Technology. 38(6), 662-666(2016).
牛连强, 陈向震, 张胜男 等. 深度连续卷积神经网络模型构建与性能分析. 沈阳工业大学学报. 38(6), 662-666(2016).

【9】Cui R X, Liu M Y and Liu M H. Facial expression recognition based on ensemble of mulitple CNNs. ∥ You Z S, Zhou J, Wang Y H, et al. Biometric recognition. Lecture notes in computer science. Cham: Springer. 9967, 511-518(2016).

【10】Goodfellow I J, Le Q V, Saxe A M et al. Measuring invariances in deep networks. [C]∥Proceedings of the 22nd International Conference on Neural Information Processing Systems, December 7-10, 2009, Vancouver, British Columbia, Canada. USA: Curran Associates Inc. 646-654(2009).

【11】Gu J X, Wang Z H, Kuen J et al. Recent advances in convolutional neural networks. Pattern Recognition. 77, 354-377(2018).

【12】Zeiler M D, Krishnan D, Taylor G W et al. Deconvolutional networks. [C]∥2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June 13-18, 2010, San Francisco, CA, USA. New York: IEEE. 2528-2535(2010).

【13】Wang M, Hao J, Yao C H et al. Sign language semantic recognition based on optimized fully convolutional neural network. Laser & Optoelectronics Progress. 55(11), (2018).
王民, 郝静, 要趁红 等. 基于优化全卷积神经网络的手语语义识别. 激光与光电子学进展. 55(11), (2018).

【14】Ivanovsky L, Khryashchev V, Lebedev A et al. Facial expression recognition algorithm based on deep convolution neural network. [C]∥2017 21st Conference of Open Innovations Association (FRUCT), November 6-10, 2017, Helsinki,Finland. New York: IEEE. 141-147(2018).

【15】Chen H J, Wang Q Q, Yang G W et al. SSD object detection algorithm with multi-scale convolution feature fusion. Journal of Frontiers of Computer Science and Technology. 13(6), 1049-1061(2019).
陈幻杰, 王琦琦, 杨国威 等. 多尺度卷积特征融合的SSD目标检测算法. 计算机科学与探索. 13(6), 1049-1061(2019).

【16】He K M, Zhang X Y, Ren S Q et al. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence. 37(9), 1904-1916(2015).

【17】Lee S H, Chan C S, Mayo S J et al. How deep learning extracts and learns leaf features for plant classification. Pattern Recognition. 71, 1-13(2017).

【18】Iwahori Y, Takada Y, Shiina T et al. Defect classification of electronic board using dense SIFT and CNN. Procedia Computer Science. 126, 1673-1682(2018).

引用该论文

Peng Lu, Peiqi Zou, Guoliang Zou. Typhoon Classification Model Based on Multi-Scale Convolution Feature Fusion[J]. Laser & Optoelectronics Progress, 2019, 56(16): 160101

卢鹏, 邹佩岐, 邹国良. 基于多尺度卷积特征融合的台风等级分类模型[J]. 激光与光电子学进展, 2019, 56(16): 160101

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF