基于CNN的中国绘画图像分类被引量：4

Classification of Chinese Painting Image Based on CNN

下载PDF

导出

摘要提出了一种基于卷积神经网络CNN的中国绘画图像分类方法.首先针对过拟合问题,提出了一种改进的合成少数类过采样技术SMOTE扩增数据,然后将扩增后的数据直接输入到CNN中,经过隐藏层的卷积和亚采样,并采用校正线性单元ReLu、S形生长曲线Sigmoid替代传统的Sigmoid激活函数,提取的数据能更好地表示其特征.实验结果表明,与传统分类方法相比,新提出的方法在中国绘画图像分类上具有良好的分类能力. This paper presented a method which based on convolutional neural network（CNN）algorithm to classify Chinese painting images.Data was augmented with an improved synthetic minority over-sampling technique（SMOTE）to avoid over fitting problem firstly.And then the augmented data was inputted to the CNN system directly.After hidden layers of convolution and sub-sampling,activation function which was replaced from traditional Sigmoid to the rectified linear units,Sigmoid.Finally,features which can be better represents data were extracted.Experimental results show that the presented method has excellent classification ability for the classification of Chinese painting image.

作者黄雪丽王小华杨冰姚金良

机构地区杭州电子科技大学计算机学院中国计量学院计算机学院

出处《杭州电子科技大学学报（自然科学版）》 2017年第2期45-50,共6页 Journal of Hangzhou Dianzi University：Natural Sciences

基金国家自然科学基金资助项目(61402143 61202280) 浙江省自然科学基金资助项目(LQ14F020012)

关键词 SMOTE ReLu+Sigmoid 卷积神经网络中国绘画图像分类 synthetic minority over-sampling technique ReLu ＋ Sigmoid convolutional neural network classification of Chinese painting

分类号 TP391 [自动化与计算机技术—计算机应用技术]

引文网络
相关文献

参考文献4

1潘卫国,鲍泓,何宁.一种传统中国书画图像的二分类方法[J].计算机科学,2012,39(3):256-259. 被引量：4
2胡正平,陈俊岭,王蒙,赵淑欢.卷积神经网络分类模型在模式识别中的新进展[J].燕山大学学报,2015,39(4):283-291. 被引量：31
3张莉,郭军.基于边界样本的训练样本选择方法[J].北京邮电大学学报,2006,29(4):77-80. 被引量：16
4王振,高茂庭.基于卷积神经网络的图像识别算法设计与实现[J].现代计算机（中旬刊）,2015(7):61-66. 被引量：49

二级参考文献34

1蒋树强,杜军,黄庆明,黄铁军,高文.Visual Ontology Construction for Digitized Art Image Retrieval[J].Journal of Computer Science & Technology,2005,20(6):855-860. 被引量：7
2Wilson D R, Martinez T R. Instance pruning techniques [C]// Proceedings of the 14th International Conference. San Francisco: Morgan Kaufmann Publishers Inc, 1997:404-411.
3Astrahan M M. Speech analysis by clustering, or the hyper-phoneme method [R]. Calif: Stanford Univ, 1970.
4Mitra P, Pal S K. Density-based multiscale data condensation [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(6): 734-747.
5Ng W W Y, Yeung D S, Cloete I. Input sample selection for rbf neural network classification problems using sensitivity measure [C]// IEEE International Conference on Systems Man and Cybernetics. Washington: [s. n.], 2003: 2593-2598.
6Tambouratzis T. Counter-clustering for training pattern selection [J]. The Computer Journal, 2000, 43 (3) :177-190.
7Lyhyaoui A, Ynez M M, Mora I. Sample selection via clustering to construct support vector-like classifiers [J]. IEEE Transactions on Neural Networks, 1999, 10 (6) :1474-1480.
8Brighton H, Mellish C. Advances in instance selection for instance-based learning algorithms [J]. Data Mining and Knowledge Discovery, 2002, 6(2): 153-172.
9Luo Dingsheng, Chen Ke. Refine decision boundaries of a statistical ensemble by active learning [C] // International Joint Conference on Neural Networks. Portland: [s.n.], 2003: 1523-1528.
10Alex Krizhevsky, Ilya Sutskever, Geoff Hinton. Imagenet classification with deep con-volutional neural networks[J]. Advances in Neural Information Processing Systems 25, 2012:1106-1114.