基于欧氏距离双比特嵌入哈希的图像检索

doi:10.3969/j.issn.0255-8297.2017.02.006

应用科学学报 ›› 2017, Vol. 35 ›› Issue (2): 193-206.doi: 10.3969/j.issn.0255-8297.2017.02.006

基于欧氏距离双比特嵌入哈希的图像检索

李蕾^1,2, 岑翼刚^1,2, 赵瑞珍^1,2, 崔丽鸿³, 王艳红^1,2

1. 北京交通大学信息科学研究所, 北京 100044;
2. 北京交通大学现代信息科学与网络技术重点实验室, 北京 100044;
3. 北京化工大学理学院, 北京 100029

收稿日期:2016-08-20 修回日期:2016-10-18 出版日期:2017-03-30 发布日期:2017-03-30
通信作者: 岑翼刚,教授,博导,研究方向:小波变换、图像信号处理、压缩感知、稀疏表示,E-mail:ygcen@bjtu.edu.cn E-mail:ygcen@bjtu.edu.cn
基金资助:
国家“863”高技术研究发展计划基金（No.2014AA015202）；国家自然科学基金（No.61272028，No.61572067）；北京市自然基金（No.4162050）；广东省自然科学基金（No.2016A030313708）资助

Euclidean Double Bits Embedding Hashing for Image Retrieval

LI Lei^1,2, CEN Yi-gang^1,2, ZHAO Rui-zhen^1,2, CUI Li-hong³, WANG Yan-hong^1,2

1. Institute of Information Science, Beijing Jiaotong University, Beijing 100044, China;
2. Key Laboratory of Information Science and Network Technology, Beijing Jiaotong University Beijing 100044, China;
3. College of Sciences, Beijing University of Chemical Technology, Beijing 100029, China

Received:2016-08-20 Revised:2016-10-18 Online:2017-03-30 Published:2017-03-30

摘要/Abstract

摘要：

提出一种基于欧氏距离的双比特嵌入哈希算法，以欧氏距离来度量二进制哈希编码之间的相似性.该方法可更好地保持原始特征空间的相似性关系，提高检索精度.另外，为了提高欧氏距离的计算速度，利用位操作实现二进制哈希编码欧氏距离的计算.对于64位的双比特嵌入哈希码，所提算法比传统欧氏距离的计算速度快400倍左右.在3个主流图像库上进行图像检索实验，与当前主流量化算法相比，该算法取得了更好的检索结果.

关键词: 图像检索, 欧氏距离, 哈希, 双比特嵌入

Abstract:

We propose a double-bit embedding hashing method based on the Euclidean distance (DBE-E). Euclidean distance is used to measure similarity between the binary hash codes to better preserve similarity relations of the original feature space and improve retrieval precision. To speed computation, bit operation is used to calculate the Euclidean distance between the hash codes. It is 400 times faster than the traditional calculation method of the Euclidean distance for double-bit embedding of 64-bit hash code. Experiments on three image data sets show that the proposed method produces better results than other popular quantization strategies of hashing.

Key words: hashing, double-bit embedding, image retrieval, Euclidean distance

中图分类号:

TP391

李蕾, 岑翼刚, 赵瑞珍, 崔丽鸿, 王艳红. 基于欧氏距离双比特嵌入哈希的图像检索[J]. 应用科学学报, 2017, 35(2): 193-206.

LI Lei, CEN Yi-gang, ZHAO Rui-zhen, CUI Li-hong, WANG Yan-hong. Euclidean Double Bits Embedding Hashing for Image Retrieval[J]. Journal of Applied Sciences, 2017, 35(2): 193-206.

参考文献

[1] 许相莉,张利彪,于哲舟,周春光. 多粒度颜色特征在图像检索中的应用(英)[J]. 应用科学学报,2009, 27(1):56-61. Xu X L, Zhang L B, Yu Z Z, Zhou C G. Application of multi-granularity color features in image retrieval[J]. Journal of Applied Sciences, 2009, 27(1):56-61.
[2] Castanon G, Saligrama V, Caron A L, Jodoin P. Real-time activity search of surveillance video[C]//IEEE 9th International Conference on Advanced Video & Signal-based Surveillance. IEEE, 2012:246-251.
[3] 赵永威, 李弼程, 高毫林. 一种基于精确欧氏位置敏感哈希的目标检索方法[J]. 应用科学学报, 2012, 30(4):349-355. Zhao Y W, Li B C, Gao H L. Object retrieval based on exact euclidean locality sensitive hashing[J]. Journal of Applied Sciences, 2012, 30(4):349-355. (in Chinese)
[4] 郑永军,张连海. 基于动态匹配词格检索的关键词检测[J]. 应用科学学报,2014, 32(2):149-155. Zheng Y J, Zhang L H. Keyword detection based on dynamic match lattice spotting[J]. Journal of Applied Sciences, 2014, 32(2):149-155. (in Chinese)
[5] Buhler J. Efficient large-scale sequence comparison by locality-sensitive hashing[J]. Bioinformatics, 2001, 17(5):419-28.
[6] Indyk P, Motwani R. Approximate nearest neighbors:towards removing the curse of dimensionality[J]. Theory of Computing, 2000, 8(11):604-613.
[7] 张一凡,余小清,安炫东,万旺根. 一种基于CUDA的局部敏感哈希算法[J]. 应用科学学报,2015, 33(5):550-558. Zhang Y F, Yu X Q, An X D, Wan W G. A locality sensitive hashing algorithm based on CUDA[J]. Journal of Applied Sciences, 2015, 33(5):550-558. (in Chinese)
[8] Raginsky M, Lazebnik S. Locality-sensitive binary codes from shift-invariant kernels[C]//Advances in Neural Information Processing Systems 22:23rd Annual Conference on Neural Information Processing Systems, 2009:1509-1517.
[9] Rahimi A, Recht B. Random features for large-scale kernel machines[J]. Advances in Neural Information Processing Systems, 2007, 20:1177-1184.
[10] Gordo A, Perronnin F, Gong Y C, Lazebnik S. Asymmetric distances for binary embeddings[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2013, 36(1):729-736.
[11] Gong Y C, Lazebnik S, Gordo A, Perronnin F. Iterative quantization:a procrustean approach to learning binary codes for large-scale image retrieval[C]//Proceedings of the 2011 IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2011:2916-2929.
[12] Weiss Y, Torralba A, Fergus R. Spectral hashing[J]. Advances in Neural Information Processing Systems, 2008, 282(3):1753-1760.
[13] Norouzi M, Fleet D J. Cartesian K-means[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition. IEEE, 2013:3017-3024.
[14] Lv Y M, Ng W W Y, Zeng Z Q, Yeung D S, Chan P P K. Asymmetric cyclical hashing for large scale image retrieval[J]. IEEE Transactions on Multimedia, 2015, 17(8):1225-1235.
[15] Kong W, Li W J. Double-bit quantization for hashing[C]//AAAI Conference on Artificial Intelligence, 2004:137-138.
[16] Liu W, Wang J, Kumar S, Chang S F. Hashing with Graphs[C]//Proceedings of the 28th International Conference on Machine Learning, 2011:1-8.
[17] Lee Y, Heo J P, Yoon S E. Quadra-Embedding:binary code embedding with low quantization error[C]//Asian Conference on Computer Vision. Springer-Verlag, 2012:214-222.
[18] Kong W, Li W J, Guo M. Manhattan hashing for large-scale image retrieval[C]//Proceedings of the 35th international ACM SIGIR conference on Research and development in information retrieval. ACM, 2012:45-54.
[19] Oliva A, Torralba A. Modeling the shape of the scene:a holistic representation of the spatial envelope[J]. International Journal of Computer Vision, 2001, 42(3):145-175.

基于欧氏距离双比特嵌入哈希的图像检索

Euclidean Double Bits Embedding Hashing for Image Retrieval

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 14

编辑推荐

Metrics

本文评价

[1]	佘维, 陈建森, 顾志豪, 田钊, 徐力, 刘炜. 基于区块链的物联网节点位置隐私保护模型[J]. 应用科学学报, 2020, 38(1): 139-151.
[2]	韦秋含, 梁海华, 张新鹏. 基于动态BoW模型的密文JPEG图像检索[J]. 应用科学学报, 2018, 36(4): 628-634.
[3]	李新伟, 夏秀珍. 基于内容结构图的鲁棒图像哈希[J]. 应用科学学报, 2016, 34(6): 691-701.
[4]	蒋翠玲, 林家骏. 一种基于遗传算法和BP网络的鲁棒图像哈希方法[J]. 应用科学学报, 2016, 34(5): 537-546.
[5]	刘凯, 唐振军, 张显全, 俞春强. 联合压缩感知和颜色向量角的彩色图像哈希方法[J]. 应用科学学报, 2015, 33(6): 595-603.
[6]	张一凡, 余小清, 安炫东, 万旺根. 一种基于CUDA的局部敏感哈希算法[J]. 应用科学学报, 2015, 33(5): 550-558.
[7]	古强1，许小东1，蒋轶2，戴旭初1. 适合传输三路数据流的最小距离MIMO 预编码[J]. 应用科学学报, 2013, 31(2): 135-141.
[8]	赵永威，李弼程，高毫林. 一种基于精确欧氏位置敏感哈希的目标检索方法[J]. 应用科学学报, 2012, 30(4): 349-355.
[9]	高毫林，李弼程，彭天强. 基于稳健哈希签名的视频流特定人物实时检测方法[J]. 应用科学学报, 2011, 29(6): 619-625.
[10]	柳培忠1;3，王守觉1;2. 利用多维空间同源连续性的图像检索[J]. 应用科学学报, 2011, 29(2): 153-158.
[11]	徐强1，孙乐昌1，张旻2，刘海涛3. Kademlia协议中的路由表自适应调节机制[J]. 应用科学学报, 2011, 29(1): 66-72.
[12]	许相莉;张利彪;于哲舟;周春光. 多粒度颜色特征在图像检索中的应用 (英)[J]. 应用科学学报, 2009, 27(1): 56-61 .
[13]	马煜;顾晓东;汪源源. 基于平均窗口平移的直方图相似性度量[J]. 应用科学学报, 2008, 26(1): 28-28 .
[14]	张二虎, 张绪进, 张志刚. 小波变换域中图像字符的定位提取方法[J]. 应用科学学报, 2006, 24(2): 135-139.