基于机器学习的图案分类研究进展

doi:10.11996/JG.j.2095-302X.2023030415

摘要/Abstract

摘要：

回顾国内外结合机器学习进行图案分类的有关研究，从研究方法、文献数据分析、图案分类以及机器学习应用层面进行系统的梳理，了解目前国内外的研究现状以及研究进展，总结图案分类的方法以及不足之处，并对今后的发展进行展望，为更深入的探索提供参考。以大量文献研究为基础，运用CiteSpace软件分析当前的研究热点及趋势，并详细对数据集构建、数据处理、特征提取和图案分类所使用的方法、经典模型以及机器学习在图案分类中的运用进行分析和总结。图案分类研究有从传统人工分类向机器分类转变的趋势；准确高效的获取目标图案特征可有效改善分类效果；图案分类研究存在数据库匮乏问题，不利于对图案分类的深入研究。通过智能化分类方法构建系统的图案数据库；并将分类技术运用于服务系统平台，实现图案活态化传承；在分类的基础上，结合各种组合算法进行图案的创新设计，推动图案的进一步发展。

关键词: CiteSpace, 可视化分析, 研究热点与趋势, 图案分类, 机器学习

Abstract:

This paper presented a review of the research on pattern classification combined with machine learning at home and abroad. The study systematically sorted out the research methods, literature data analysis, pattern classification and machine learning applications, aiming to understand the current research status and research progress at home and abroad. This paper also summarized the methods and shortcomings of pattern classification and looked forward to the future development, providing reference for further exploration. Based on a large number of literature studies, CiteSpace software was employed to analyze the current research hotspots and trends. The methods, classical models, and the application of machine learning in pattern classification used in dataset construction, data processing, feature extraction, and pattern classification were analyzed and summarized in detail. The pattern classification research had a tendency of evolving from traditional manual classification to machine classification. The accurate and efficient acquisition of target pattern features can effectively improve the classification effect. There is a lack of databases in pattern classification research, which hinders the in-depth study of pattern classification. The systematic pattern database can be constructed by intelligent classification methods. The classification technology can be applied to the service system platform to realize the living inheritance of patterns. On the basis of classification, combined with various combination algorithms, innovative design of patterns can be carried out to promote the further development of patterns.

Key words: CiteSpace, visual analysis, research hotspots and trends, pattern classification, machine learning

中图分类号:

TB472

边坤, 梁慧. 基于机器学习的图案分类研究进展[J]. 图学学报, 2023, 44(3): 415-426.

BIAN Kun, LIANG Hui. Research progress of pattern classification based on machine learning[J]. Journal of Graphics, 2023, 44(3): 415-426.

图/表 12

参考文献 46

[1]	刘则渊, 陈悦, 侯海燕. 科学知识图谱: 方法与应用[M]. 北京: 人民出版社, 2008: 1.
	LIU Z Y, CHEN Y, HOU H Y. Mapping knowledge domains: methods and application[M]. Beijing: People's Publishing House, 2008: 1. (in Chinese)
[2]	何晓萍, 黄龙. 大数据领域演进路径、研究热点与前沿的可视化分析[J]. 现代情报, 2015, 35(4): 46-51. DOI
	HE X P, HUANG L. Visualization analysis of evolution path, research hotspots and frontiers of big data[J]. Journal of Modern Information, 2015, 35(4): 46-51. (in Chinese)
[3]	张宗登, 游天梅, 张红颖. 中国苗族服饰研究热点及发展趋势: 基于CNKI数据的可视化研究[J]. 湖南包装, 2022, 37(1): 22-26, 36.
	ZHANG Z D, YOU T M, ZHANG H Y. Research hotspots and development trends of Chinese Miao costumes: based on CiteSpace visual analysis[J]. Hunan Packaging, 2022, 37(1): 22-26, 36. (in Chinese)
[4]	李杰, 陈超美. CiteSpace: 科技文本挖掘及可视化[M]. 北京: 首都经济贸易大学出版社, 2008: 93.
	LI J, CHEN C M. CiteSpace: scientific text mining and visualization[M]. Beijing: Capital University of Economics and Business Press, 2008: 93.
[5]	殷燕, 刘军平. 国内副文本研究三十年(1986—2016): 基于CiteSpace的科学计量分析[J]. 上海翻译, 2017(4): 22-26, 94.
	YIN Y, LIU J P. Paratextual studies in China(1986—2016): a scientometric analysis by using CiteSpace[J]. Shanghai Journal of Translators, 2017(4): 22-26, 94. (in Chinese)
[6]	张志宇. 基于深度学习的民族图案的识别[D]. 长春: 吉林大学, 2019.
	ZHANG Z Y. Recognition of national patterns based on deep learning[D]. Changchun: Jilin University, 2019. (in Chinese)
[7]	CHU W T, WU Y L. Deep correlation features for image style classification[C]// The 24th ACM International Conference on Multimedia. New York: ACM, 2016: 402-406.
[8]	周惠怡, 刘颂凯, 张磊, 等. 考虑误分类约束的电力系统暂态稳定评估[J]. 电力系统及其自动化学报, 2022, 34(6): 71-78.
	ZHOU H Y, LIU S K, ZHANG L, et al. Transient stability assessment for power system considering misclassification constraints[J]. Proceedings of the CSU-EPSA, 2022, 34(6): 71-78. (in Chinese)
[9]	杨秀芹, 张华熊. 双核压缩激活神经网络艺术图像分类[J]. 中国图象图形学报, 2020, 25(5): 967-976.
	YANG X Q, ZHANG H X. Art image classification with double kernel squeeze-and-excitation neural network[J]. Journal of Image and Graphics, 2020, 25(5): 967-976. (in Chinese)
[10]	杨冰. 基于艺术风格的绘画图像分类研究[D]. 杭州: 浙江大学, 2013.
	YANG B. Research on painting image classification based on aesthetic style[D]. Hangzhou: Zhejiang University, 2013. (in Chinese)
[11]	周长敏, 佘佐明, 吴安丽. 基于卷积神经网络的蜡染纹样分类[J]. 现代计算机, 2021(22): 117-121.
	ZHOU C M, SHE Z M, WU A L. Classification of indigo batik pattern based on convolutional neural networks[J]. Modern Computer, 2021(22): 117-121. (in Chinese)
[12]	AZHAR R, TUWOHINGIDE D, KAMUDI D, et al. Batik image classification using SIFT feature extraction, bag of features and support vector machine[J]. Procedia Computer Science, 2015, 72: 24-30. DOI URL
[13]	周长春. 美术作品检索与分类研究[D]. 杭州: 浙江大学, 2015.
	ZHOU C C. Art works retrieval and classification[D]. Hangzhou: Zhejiang University, 2015. (in Chinese)
[14]	黎克, 钱文华, 王成学, 等. 基于图神经网络的东巴画小样本分类[J]. 计算机辅助设计与图形学学报, 2021, 33(7): 1073-1083.
	LI K, QIAN W H, WANG C X, et al. Dongba painting few-shot classification based on graph neural network[J]. Journal of Computer-Aided Design & Computer Graphics, 2021, 33(7): 1073-1083. (in Chinese)
[15]	钱文华, 徐丹, 徐瑾, 等. 基于信息熵的风格绘画分类研究[J]. 图学学报, 2019, 40(6): 991-999.
	QIAN W H, XU D, XU J, et al. Artistic paintings classification based on information entropy[J]. Journal of Graphics, 2019, 40(6): 991-999. (in Chinese)
[16]	JIANG S Q, HUANG Q M, YE Q X, et al. An effective method to detect and categorize digitized traditional Chinese paintings[J]. Pattern Recognition Letters, 2006, 27(7): 734-746. DOI URL
[17]	CONDOROVICI R, FLOREA C, VRÂNCEANU R, et al. Perceptually-inspired artistic genre identification system in digitized painting collections[C]// The 18th Scandinavian Conference on Image Analysis. Cham: Springer International Publishin, 2013: 687-696.
[18]	GILCHRIST A, KOSSYFIDIS C, BONATO F, et al. An anchoring theory of lightness perception[J]. Psychological Review, 1999, 106(4): 795-834. PMID
[19]	KUMAR P, YILDIRIM E A. Minimum-volume enclosing ellipsoids and core sets[J]. Journal of Optimization Theory and Applications, 2005, 126(1): 1-21. DOI URL
[20]	GRIGORESCU S E, PETKOV N, KRUIZINGA P. Comparison of texture features based on Gabor filters[J]. IEEE Transactions on Image Processing: a Publication of the IEEE Signal Processing Society, 2002, 11(10): 1160-1167. DOI URL
[21]	王子翔. 基于VGG19的迪庆藏族装饰图案特征提取系统设计与应用[D]. 昆明: 昆明理工大学, 2021.
	WANG Z X. Design and application of Diqing Tibetan decorative pattern feature extraction system based on VGG19[D]. Kunming: Kunming University of Science and Technology, 2021. (in Chinese)
[22]	白茹意, 郭小英, 贾春花. 基于特征融合的小样本抽象画图像情感预测[J]. 计算机应用, 2020, 40(8): 2207-2213. DOI
	BAI R Y, GUO X Y, JIA C H. Sentiment prediction of small sample abstract painting image based on feature fusion[J]. Journal of Computer Applications, 2020, 40(8): 2207-2213. (in Chinese) DOI
[23]	ZHENG G, WANG B Z, SUN Y C, et al. Study on the traditional pattern retrieval method of minorities in Gansu Province[J]. IOP Conference Series: Materials Science and Engineering, 2018, 322: 052049. DOI URL
[24]	杨秀芹. 基于深度学习的艺术图像分类算法研究[D]. 杭州: 浙江理工大学, 2020.
	YANG X Q. Research on the art image classification algorithm based on deep learning[D]. Hangzhou: Zhejiang Sci-Tech University, 2020. (in Chinese)
[25]	范涛, 王昊, 李跃艳, 等. 基于多模态融合的非遗图片分类研究[J]. 数据分析与知识发现, 2022, 6(S1): 329-337.
	FAN T, WANG H, LI Y Y, et al. Classifying images of intangible cultural heritages with multimodal fusion[J]. Data Analysis and Knowledge Discovery, 2022, 6(S1): 329-337. (in Chinese)
[26]	KARNEWAR A, KANAWADAY A, SAWANT C, et al. Classification of abstract images using machine learning[C]// 2017 International Conference on Deep Learning Technologies. New York: ACM, 2017: 36-40.
[27]	孔谦, 史卓, 冯业, 等. 瑶族纹样符号的智能分类方法[J]. 包装工程, 2021, 42(10): 244-250.
	KONG Q, SHI Z, FENG Y, et al. The intelligent classification method of Yao patterns[J]. Packaging Engineering, 2021, 42(10): 244-250. (in Chinese)
[28]	陈玉红, 刘晓静. 基于卷积神经网络的唐卡尊像自动分类研究[J]. 计算机技术与发展, 2021, 31(12): 167-174.
	CHEN Y H, LIU X J. Research on automatic classification of thangka images based on convolutional neural network[J]. Computer Technology and Development, 2021, 31(12): 167-174. (in Chinese)
[29]	WANG T J, WANG W L. Research on a thangka image classification method based on support vector machine[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2019, 33(12): 1954030. DOI URL
[30]	YANG Z H. Classification of picture art style based on VGGNET[J]. Journal of Physics: Conference Series, 2021, 1774(1): 012043. DOI
[31]	盛家川, 陈雅琦, 王君, 等. 融合人类认知网络优化的中国画情感识别[J]. 模式识别与人工智能, 2020, 33(2): 141-149. DOI
	SHENG J C, CHEN Y Q, WANG J, et al. Chinese paintings sentiment recognition via CNN optimization with human cognition[J]. Pattern Recognition and Artificial Intelligence, 2020, 33(2): 141-149. (in Chinese) DOI
[32]	潘森垒, 钱文华, 曹进德, 等. 基于注意力机制的东巴画情感分类[J]. 图学学报, 2023, 44(1): 59-66.
	PAN S L, QIAN W H, CAO J D, et al. Learning attention for Dongba paintings emotion classification[J]. Journal of Graphics, 2023, 44(1): 59-66. (in Chinese)
[33]	CETINIC E, LIPIC T, GRGIC S. Fine-tuning convolutional neural networks for fine art classification[J]. Expert Systems With Applications, 2018, 114: 107-118. DOI URL
[34]	HIARY H, SAADEH H, SAADEH M, et al. Flower classification using deep convolutional neural networks[J]. IET Computer Vision, 2018, 12(6): 855-862. DOI URL
[35]	谭晓明. 图案艺术中“形”与中国绘画中“神”之刍析[J]. 文化学刊, 2018(2): 220-223.
	TAN X M. On “form” in pattern art and “God” in China’s painting[J]. Culture Journal, 2018(2): 220-223. (in Chinese)
[36]	高友文, 周本君, 胡晓飞. 基于数据增强的卷积神经网络图像识别研究[J]. 计算机技术与发展, 2018, 28(8): 62-65.
	GAO Y W, ZHOU B J, HU X F. Research on image recognition of convolution neural network based on data enhancement[J]. Computer Technology and Development, 2018, 28(8): 62-65. (in Chinese)
[37]	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. [2022-05-29]. https://arxiv.org/abs/1409.1556.
[38]	SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2015: 1-9.
[39]	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 770-778.
[40]	HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 2261-2269.
[41]	陆健强, 林佳翰, 黄仲强, 等. 基于Mixup算法和卷积神经网络的柑橘黄龙病果实识别研究[J]. 华南农业大学学报, 2021, 42(3): 94-101.
	LU J Q, LIN J H, HUANG Z Q, et al. Identification of citrus fruit infected with Huanglongbing based on Mixup algorithm and convolutional neural network[J]. Journal of South China Agricultural University, 2021, 42(3): 94-101. (in Chinese)
[42]	赵含笑. 基于深度学习的刺绣图像分类与识别研究[D]. 西宁: 青海师范大学, 2020.
	ZHAO H X. Classification and recognition algorithm of embroidery images based on deep learning[D]. Xining: Qinghai Normal University, 2020. (in Chinese)
[43]	贾小军, 邓洪涛, 刘子豪, 等. 基于VGGNet卷积神经网络的蓝印花布纹样分类[J]. 光电子·激光, 2019, 30(8): 867-875.
	JIA X J, DENG H T, LIU Z H, et al. Vein pattern classification based on VGGNet convolutional neural network for blue calico[J]. Journal of Optoelectronics· Laser, 2019, 30(8): 867-875. (in Chinese)
[44]	贾小军, 叶利华, 邓洪涛, 等. 基于卷积神经网络的蓝印花布纹样分类[J]. 纺织学报, 2020, 41(1): 110-117.
	JIA X J, YE L H, DENG H T, et al. Elements classification of vein patterns using convolutional neural networks for blue calico[J]. Journal of Textile Research, 2020, 41(1): 110-117. (in Chinese)
[45]	KELEK M O, CALIK N, YILDIRIM T. Painter classification over the novel art painting data set via the latest deep neural networks[J]. Procedia Computer Science, 2019, 154: 369-376. DOI URL
[46]	YANG H, MIN K. Classification of basic artistic media based on a deep convolutional approach[J]. The Visual Computer, 2020, 36(3): 559-578. DOI

序号	出现频次	关键词	中介中心性	出现年
1	64	深度学习	0.40	2007
2	42	迁移学习	0.12	1990
3	14	特征提取	0.19	2008
4	6	特征融合	0.11	2008
5	9	神经网络	0.09	2007

序号	出现频次	关键词	中介中心性	出现年
1	64	深度学习	0.40	2007
2	42	迁移学习	0.12	1990
3	14	特征提取	0.19	2008
4	6	特征融合	0.11	2008
5	9	神经网络	0.09	2007

经典网络模型	时间(年)	优势	不足
AlexNet	2012	以ReLu激活函数替代Sigmoid函数，避免梯度消失，加快收敛速度， Dropout降低过拟合，泛化能力增强	对所需训练的数据量需求较大，易造成计算量大的问题、结构层深度增加易引起过拟合、梯度消失或爆炸
VGGNet	2014	使用小卷积核代替大卷积核，网络模型结构更深、更宽，表现效果更好	相比AlexNet需要更大的数据量，训练耗费时间较长、深度增加易引起过拟合、梯度消失或爆炸
GoogleNet	2015	Inception结构使模型加深、加宽的同时稀疏连接可减少参数以及计算量、密集矩阵实现高计算性能	网络深度增加不利于模型的收敛
ResNet	2016	残差连接使模型结构更深、性能更强更高效；通道数增加利于高级特征的获取	特征过度重用，特征冗余
DenseNet	2017	拓展网络连接、信息流最大化、增强信息传播、参数量减少、支持特征重用	训练耗费时间较长，在小样本情况下容易过拟合

经典网络模型	时间(年)	优势	不足
AlexNet	2012	以ReLu激活函数替代Sigmoid函数，避免梯度消失，加快收敛速度， Dropout降低过拟合，泛化能力增强	对所需训练的数据量需求较大，易造成计算量大的问题、结构层深度增加易引起过拟合、梯度消失或爆炸
VGGNet	2014	使用小卷积核代替大卷积核，网络模型结构更深、更宽，表现效果更好	相比AlexNet需要更大的数据量，训练耗费时间较长、深度增加易引起过拟合、梯度消失或爆炸
GoogleNet	2015	Inception结构使模型加深、加宽的同时稀疏连接可减少参数以及计算量、密集矩阵实现高计算性能	网络深度增加不利于模型的收敛
ResNet	2016	残差连接使模型结构更深、性能更强更高效；通道数增加利于高级特征的获取	特征过度重用，特征冗余
DenseNet	2017	拓展网络连接、信息流最大化、增强信息传播、参数量减少、支持特征重用	训练耗费时间较长，在小样本情况下容易过拟合