Please wait a minute...
Advanced Search
数据分析与知识发现  2018, Vol. 2 Issue (3): 49-59     https://doi.org/10.11925/infotech.2096-3467.2017.1023
  研究论文 本期目录 | 过刊浏览 | 高级检索 |
基于图像特征匹配的推荐模型研究*
刘东苏, 霍辰辉()
西安电子科技大学经济与管理学院 西安 710071
Recommending Image Based on Feature Matching
Liu Dongsu, Huo Chenhui()
School of Economics and Management, Xidian University, Xi’an 710071, China
全文: PDF (1918 KB)   HTML ( 1
输出: BibTeX | EndNote (RIS)      
摘要 

目的】基于LSH算法将图像匹配应用到图像推荐模型中, 与传统推荐模型结合, 提高推荐结果准确度。【方法】提取图像SIFT特征作为图像匹配标准, 改进基于p-Stable Distribution的LSH算法, 实现高维度下大量图片的搜索匹配, 最后融合现有协同过滤算法提出ICF-LSH推荐算法构建融合推荐模型, 并采用Python语言予以实现。【结果】使用不同的数据集对本文提出的算法进行验证, 实验表明改进的LSH算法对召回率和错误率都有一定的优化, 通过匹配耗时和Hash表长度可知该算法优化了内存利用和搜索匹配效率。由融合推荐模型的平均绝对误差MAE和精确度Precision可知, 相对传统的协同过滤推荐算法, 本文提出的ICF-LSH推荐算法提高了推荐结果的精准度。【局限】在提取图像特征时仅使用SIFT特征, 后续研究中可以尝试使用多种图像特征作为匹配依据, 提高匹配结果的可靠性。【结论】图像匹配算法基于LSH进行了一定改进, 提高了图像相似度匹配的效率, 此外, 本文提出的融合推荐模型能显著提升推荐效果。

服务
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章
刘东苏
霍辰辉
关键词 SIFT特征LSH图像匹配推荐系统    
Abstract

[Objective] This paper developed an image recommendation model based on feature matching technique and the LSH algorithm, aiming to improve the accuracy of recommendations. [Methods] First, we extracted the image’s SIFT features as the matching criteria. Then, we modified the LSH algorithm to retrieve images in high dimensional settings. Finally, we proposed an ICF-LSH algorithm based on the collaborative filtering techniques to build fusion recommendation model. [Results] We examined the proposed algorithm with various datasets and achieved better recall and precision rates for image recommendation. [Limitations] Only used the SIFT feature to extract image features. More research is needed to explore other matching features. [Conclusions] The proposed model improves the performance of image matching and recommendation systems.

Key wordsSIFT Feature    LSH    Image Matching    Recommendation System
收稿日期: 2017-10-13      出版日期: 2018-04-03
ZTFLH:  N99 TP391  
基金资助:*本文系国家自然科学基金项目“基于可信语义Wiki的知识库构建方法与研究应用”(项目编号:71203173)和国家自然科学青年基金项目“大规模动态社交网络社团检测算法研究”(项目编号: 71401130)的研究成果之一
引用本文:   
刘东苏, 霍辰辉. 基于图像特征匹配的推荐模型研究*[J]. 数据分析与知识发现, 2018, 2(3): 49-59.
Liu Dongsu,Huo Chenhui. Recommending Image Based on Feature Matching. Data Analysis and Knowledge Discovery, 2018, 2(3): 49-59.
链接本文:  
http://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/10.11925/infotech.2096-3467.2017.1023      或      http://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/Y2018/V2/I3/49
  基于图像特征匹配的商品推荐模型构建流程
  图片特征点的匹配结果
  图片搜索匹配的主要工作原理
  数据集基本结构图
  Hash Table的数据结构
  推荐系统结构模型
数据集大小 召回率 错误率 匹配耗时(s) Hash表长度
2 000 0.82 1.421 0.042 12
4 000 0.90 1.337 0.069 19
6 000 0.93 1.223 0.085 34
8 000 0.96 1.069 0.126 47
  训练结果数据
  Hash表长度与召回率关系图Ⅰ
数据集大小 召回率 错误率 匹配耗时(s) Hash表长度
2 000 0.41 5.913 0.142 25
4 000 0.50 5.673 0.278 39
6 000 0.64 4.825 0.455 49
8 000 0.77 4.546 0.724 61
  不引入H1H2的训练结果数据
  Hash表长度与召回率关系图Ⅱ
  ICF算法和ICF-LSH融合推荐算法结果对比
[1] 余冲. 基于深度学习的协同过滤模型研究[D]. 深圳: 深圳大学, 2017.
[1] (Yu Chong.A Study on Collaborative Filtering Model Based on Deep Learning [D]. Shenzhen: Shenzhen University, 2017.)
[2] Krizhevsky A, Sutskever I, Hinton G E.ImageNet Classification with Deep Convolutional Neural Networks[C]// Proceedings of the 25th International Conference on Neural Information Processing Systems, Lake Tahoe, USA. Curran Associates Inc., 2012: 1097-1105.
[3] LeCun Y, Bengio Y, Hinton G. Deep Learning[J]. Nature, 2015, 521(7553): 436-444.
doi: 10.1038/nature14539
[4] 郑孝遥, 鲍煜, 孙忠宝, 等. 一种基于信任的协同过滤推荐模型[J]. 计算机工程与应用, 2016, 52(5): 50-54.
doi: 10.3778/j.issn.1002-8331.1507-0016
[4] (Zheng Xiaoyao, Bao Yu, Sun Zhongbao, et al.Collaborative Filtering Recommendation Model Based on Trust[J]. Computer Engineering and Applications, 2016, 52(5): 50-54.)
doi: 10.3778/j.issn.1002-8331.1507-0016
[5] de Campos L M, Fernández-Luna J M, Huete J F, et al. Combining Content-based and Collaborative Recommendations: A Hybrid Approach Based on Bayesian Networks[J]. International Journal of Approximate Reasoning, 2010, 51(7): 785-799.
doi: 10.1016/j.ijar.2010.04.001
[6] Melville P, Mooney R J, Nagarajan R.Content-boosted Collaborative Filtering for Improved Recommendations[C]// Proceedings of the 18th National Conference on Artificial Intelligence, Edmonton, Canada. Menlo Park, USA: American Association for Artificial Intelligence, 2002: 187-192.
[7] He R, McAuley J. VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback[C]// Proceedings of the 30th AAAI Conference on Artificial Intelligence (AAAI-16), Phoenix, USA. 2016.
[8] 苏栋梁. 融合图像相似性与协同过滤的个性化推荐算法研究[D]. 苏州: 苏州大学, 2014.
[8] (Su Dongliang.Research on Personalized Recommendation Algorithm Integrating Image Similarity and Collaborative Filtering[D]. Suzhou: Soochow University, 2014.)
[9] 朱鹏新. 基于图像内容的电商物品检索与推荐系统研究[D]. 广州: 华南理工大学, 2013.
[9] (Zhu Pengxin.Research of Content-based Commercial Product Image Retrieval and Recommendation System [D]. Guangzhou: South China University of Technology, 2013.)
[10] Indyk P.Approximate Nearest Neighbors: Towards Removing the Curse of Dimensionality[J]. Theory of Computing, 1998(11): 604-613.
[11] Szmit R.Locality Sensitive Hashing for Similarity Search Using Map Reduce on Large Scale Data[A]//Language Processing and Intelligent Information Systems[M]. Springer Berlin Heidelberg, 2013: 171-178.
[12] Andoni A, Indyk P.Near-optimal Hashing Algorithms for Approximate Nearest Neighbor in High Dimensions[C]// Proceedings of the 47th Annual IEEE Symposium on Foundations of Computer Science, Berkeley, USA. New York, USA: IEEE, 2006: 459-468.
[13] 曹玉东, 刘艳洋, 贾旭, 等. 基于改进的局部敏感哈希算法实现图像型垃圾邮件过滤[J]. 计算机应用研究, 2016, 33(6): 1693-1696.
doi: 10.3969/j.issn.1001-3695.2016.06.021
[13] (Cao Yudong, Liu Yanyang, Jia Xu, et al.Image Spam Filtering with Improved LSH Algorithm[J]. Application Research of Computers, 2016, 33(6): 1693-1696.)
doi: 10.3969/j.issn.1001-3695.2016.06.021
[14] 龚卫国, 张旋, 李正浩. 基于改进局部敏感散列算法的图像配准[J]. 光学精密工程, 2011, 19(6): 1375-1383.
doi: 10.3788/OPE.20111906.1375
[14] (Gong Weiguo, Zhang Xuan, Li Zhenghao.Image Registration Based on Extended LSH[J]. Optics and Precision Engineering, 2011, 19(6): 1375-1383.)
doi: 10.3788/OPE.20111906.1375
[15] 杨宇. 基于深度学习特征的图像推荐系统[D]. 成都: 电子科技大学, 2015.
[15] (Yang Yu.Image Recommendation System Based on the Image Features Obtained from Deep Learning[D]. Chengdu: University of Electronic Science and Technology of China, 2015.)
[16] 傅卫平, 秦川, 刘佳, 等. 基于SIFT算法的图像目标匹配与定位[J]. 仪器仪表学报, 2011, 32(1): 163-169.
[16] (Fu Weiping, Qin Chuan, Liu Jia, et al.Matching and Location of Image Object Based on SIFT Algorithm[J]. Chinese Journal of Scientific Instrument, 2011, 32(1): 163-169.)
[17] Suzuki T, Amano Y, Hashizume T.Vision Based Localization of a Small UAV for Generating a Large Mosaic Image[C]// Proceedings of SICE Annual Conference 2010, Taipei, Taiwan,China. New York, USA: IEEE, 2010: 2960-2964.
[18] 熊英, 马惠敏. 3维物体SIFT特征的提取与应用[J]. 中国图象图形学报, 2010, 15(5):814-819.
doi: 10.11834/jig.20100516
[18] (Xiong Ying, Ma Huimin.Extraction and Application of 3D Object SIFT Feature[J]. Journal of Image and Graphics, 2010, 15(5): 814-819.)
doi: 10.11834/jig.20100516
[19] Kounalakis T, Triantafyllidis G A.3D Scene’s Object Detection and Recognition Using Depth Layers and SIFT-based Machine Learning[J]. 3d Research, 2011, 2(3): 1-11.
doi: 10.1007/3DRes.03(2011)6
[20] Bai J, Ma Y, Li J, et al.Novel Averaging Window Filter for SIFT in Infrared Face Recognition[J]. Chinese Optics Letters, 2011, 9(8): 081002.
doi: 10.3788/COL
[21] Lowe D G.Distinctive Image Features from Scale-Invariant Keypoints[J]. International Journal of Computer Vision, 2004, 60(2):91-110.
doi: 10.1023/B:VISI.0000029664.99615.94
[22] 王旭乐. 基于内容的图像检索系统中高维索引技术的研究[D]. 武汉: 华中科技大学, 2008.
[22] (Wang Xule.Research on High-dimensional Indexing Technology in Content-based Image Retrieval System [D]. Wuhan: Huazhong University of Science and Technology, 2008.)
[23] 丁雪梅, 王维雅, 黄向东. 基于差分和特征不变量的运动目标检测与跟踪[J]. 光学精密工程, 2007, 15(4): 570-576.
doi: 10.3321/j.issn:1004-924X.2007.04.021
[23] (Ding Xuemei, Wang Weiya, Huang Xiangdong.New Method for Detecting and Tracking of Moving Target Based on Difference and Invariant[J]. Optics & Precision Engineering, 2007, 15(4): 570-576.)
doi: 10.3321/j.issn:1004-924X.2007.04.021
[24] Indyk P. Stable Distributions, Pseudorandom Generators, Embeddings and Data Stream Computation[C]// Proceedings of the 41st Annual Symposium on Foundations of Computer Science, Redondo Beach, USA. New York, USA: IEEE, 2000.
[25] Datar M, Immorlica N, Indyk P, et al.Locality-sensitive Hashing Scheme Based on P-stable Distributions[C]// Proceedings of the 20th Symposium on Computational Geometry, Brooklyn, USA. New York, USA: ACM, 2004: 253-262.
[26] Andoni A, Razenshteyn I.Optimal Data-Dependent Hashing for Approximate Near Neighbors[C]//Proceedings of the 47th Annual ACM Symposium on Theory of Computing, Portland, USA. New York, USA: ACM, 2015: 793-801.
[27] The MIR-Flickr Retrieval Evaluation [EB/OL]. [2017-08-17]. .
[1] 杨恒,王思丽,祝忠明,刘巍,王楠. 基于并行协同过滤算法的领域知识推荐模型研究*[J]. 数据分析与知识发现, 2020, 4(6): 15-21.
[2] 温彦,马立健,曾庆田,郭文艳. 基于地理信息偏好修正和社交关系偏好隐式分析的POI推荐 *[J]. 数据分析与知识发现, 2019, 3(8): 30-39.
[3] 焦富森,李树青. 基于物品质量和用户评分修正的协同过滤推荐算法 *[J]. 数据分析与知识发现, 2019, 3(8): 62-67.
[4] 张怡文,张臣坤,杨安桔,计成睿,岳丽华. 基于条件型游走的四部图推荐方法*[J]. 数据分析与知识发现, 2019, 3(4): 117-125.
[5] 刘丹. 利用Apache Mahout部署个性化图书推荐服务[J]. 现代图书情报技术, 2015, 31(10): 102-108.
[6] 谭学清, 何珊. 音乐个性化推荐系统研究综述[J]. 现代图书情报技术, 2014, 30(9): 22-32.
[7] 张晓燕, 张朋柱, 李嘉, 刘景方. 在线群体创新中的图片推荐方法研究[J]. 现代图书情报技术, 2014, 30(6): 94-99.
[8] 罗琳, 梁桂生, 蔡军. 基于分众分类法的图书馆书目推荐系统[J]. 现代图书情报技术, 2014, 30(4): 14-19.
[9] 姜书浩, 薛福亮. 一种利用协同过滤预测和模糊相似性改进的基于内容的推荐方法[J]. 现代图书情报技术, 2014, 30(2): 41-47.
[10] 胡新明, 罗建军, 夏火松. 基于商品领域知识的交互式推荐系统[J]. 现代图书情报技术, 2014, 30(10): 56-62.
[11] 田野, 祝忠明, 刘树栋. 基于关联数据的推荐系统综述[J]. 现代图书情报技术, 2013, 29(10): 1-7.
[12] 李嘉, 张朋柱, 李欣苗, Jihie Kim. 一种通过挖掘研讨记录来促进学生思考的在线督导系统[J]. 现代图书情报技术, 2012, 28(4): 10-16.
[13] 张慧颖, 薛福亮. 一种利用Vague集理论改进的协同过滤推荐算法[J]. 现代图书情报技术, 2012, 28(3): 35-39.
[14] 张慧颖, 薛福亮. 一种集成客户终身价值与协同过滤的推荐方法[J]. 现代图书情报技术, 2012, 28(1): 46-52.
[15] 边鹏, 赵妍, 苏玉召. 一种改进的K-means算法最佳聚类数确定方法[J]. 现代图书情报技术, 2011, 27(9): 34-40.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
版权所有 © 2015 《数据分析与知识发现》编辑部
地址:北京市海淀区中关村北四环西路33号 邮编:100190
电话/传真:(010)82626611-6626,82624938
E-mail:jishu@mail.las.ac.cn