计算机科学 ›› 2014, Vol. 41 ›› Issue (8): 281-285.doi: 10.11896/j.issn.1002-137X.2014.08.059

• 图形图像与模式识别 • 上一篇    下一篇

BoF扩展模型研究

梁晔,刘宏哲,于剑   

  1. 北京联合大学北京市信息服务工程重点实验室 北京100101;北京交通大学计算机与信息技术学院 北京100044;北京联合大学北京市信息服务工程重点实验室 北京100101;北京交通大学计算机与信息技术学院 北京100044
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受国家自然科学基金项目(61271369,61372148),北京市信息服务工程重点实验室开放课题(ZK20201402),北京联合大学“新起点”计划项目(ZK201211)资助

Research on Extended BoF Model

LIANG Ye,LIU Hong-zhe and YU Jian   

  • Online:2018-11-14 Published:2018-11-14

摘要: BoF特征是目前应用最广泛的图像表示方法。针对BoF特征编码简单、缺乏空间信息的缺点,对传统BoF流程中的特征编码和特征汇集阶段进行改进,提出了用于图像分类的新图像表示方法。首先对图像进行了基于多环划分的特征汇集的区域选择,嵌入了更多的空间信息;其次,根据密采样的特征描述子符合长尾分布的事实以及场景中特征分布比较均匀的特点,提出了适合于场景图像分类的多视觉词硬编码的编码方法。新的图像表示方法保存了BoF范式的优点,且特征表示更加紧凑、空间信息更加丰富。实验结果证明了所提方法的有效性。

关键词: 特征包,特征量化,特征汇集,图像表示,图像分类

Abstract: BoF feature is one of the most popular image representation methods by now.Aiming at the weaknesses of hard assignment coding and discarding spatial information,the improvements of feature coding and pooling in traditional BoF paradigm were proposed.The new image representation can be used for image classification.First,multi-annulus partition method was proposed for feature pooling,which can be embedded more spatial information.Second,multi-words hard assignment coding method was proposed according to long-tail distribution of dense samples and relatively even distribution of features in scene images.The new representation not only preserves merits of BoF paradigm but also is more compact and has more spatial information.The experimental results prove the efficiency of the new method.

Key words: BoF,Feature quantization,Feature pooling,Image representation,Image classification

[1] Csurka G,Dance C R,Fan Li-xin,et al.Visual categorizationwith bags of keypoints[C]∥Proceedings of European Conference Computer Vision 2004,workshop on Statistical Learning in Computer Vision.Prague,Czech Republic:Springer-Verlag LNCS,2004:59-74
[2] Coates A,Ng A Y.The importance of encodingversus trainingwith sparse coding and vector quantization[C]∥ICMA.2011
[3] Rigamonti R,Brown M A,Lepetit V.Are sparse representations really relevant for image classification[C]∥CVPR.2011
[4] Sivic J,Zisserman A.Video google:A Text Retrieval Approach to Object Matching in Videos[C]∥Proceedings of IEEE International Conference on Computer Vision.Nice,France:IEEE Computer Society,2003:1470-1477
[5] Lee H,Battle A,Raina R,et al.Efficient sparse coding algo-rithms[C]∥Proceedings of Advances in Neural Information Processing System,2006.2006:801-808
[6] Yang J,Yu K,Gong Y,et al.Linear Spatial Pyramid Matching Using Sparse Coding for Image Classification[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Miami,Florida,USA:IEEE Computer Society,2009:1794-1801
[7] Gao S,Tsang I,Chia L,et al.Local Features Are Not Lonely-Laplacian Sparse Coding for Image Classification[C]∥Procee-dings of IEEE Conference on Computer Vision and Pattern Re-cognition.San Francisco,CA,USA:IEEE Computer Society,2010:3555-3561
[8] Wang J,Yang J,Yu K,et al.Locality-constrained linear coding for image classification[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.San Francisco,CA,USA:IEEE Computer Society,2010:3360-3367
[9] Yu K,Zhang T,Gong Y.Nonlinear Learning Using Local Coordinate Coding[C]∥Proceedings of Advances in Neural Information Processing System,2009.Vancouver,British Columbia,Canada:Springer,2009
[10] Liu Ling-qiao,Wang Lei,Liu Xin-wang.In Defense of Soft-as-signment Coding[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2011.Colorado Springs,CO,USA:IEEE Computer Society,2011:2486-2493
[11] Huang Y,Huang K,Yu Y,et al.Salient Coding for Image Classification[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2011.Colorado Springs,CO,USA:IEEE Computer Society,2011:1753-1760
[12] Lazebnik S,Schmid C,Ponce J.Beyond bags of features:Spatial pyramid matching for recognizing natural scene categories[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2006.New York,NY,USA:IEEE Computer Society,2006:2169-2178
[13] Avila S,Thome N,Cord M,et al.Bossa:Extended Bow Formali-sm for Image Classification[C]∥Proceedings of International Conference on Image Processing,2011.Brussels,Belgium:IEEE Computer Society,2011:2909-2912
[14] Harada T,Ushiku Y,Yamashita Y,et al.Discriminative Spatial Pyramid[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2011.Colorado Springs,CO,USA:IEEE Computer Society,2011:1617-1624
[15] Cao Yang,Wang Chang-hu,Li Zhi-wei,et al.Spatial-Bag-of-Features[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2010.San Francisco,CA,USA:IEEE Computer Society,2010:3352-3359
[16] Malinowski M,Fritz M.Learnable Pooling Regions for ImageClassification[C]∥Proceedings of CoRR.2013
[17] Jia Yang-qing,Huang Chang,Darrell T.Beyond Spatial Pyra-mids:Receptive Field Learning for Pooled Image Features[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2012.Providence,RI,USA:IEEE Computer Society,2012:3370-3377
[18] McCann S,Lowe D G.Local Naive Bayes Nearest Neighbor for image classification[C]∥IEEE Conference on Computer Vision and Pattern Recognition.2012:3650-3656
[19] Lowe D.Distinctive Image Features from Scale-invariant Key-points[J].International Journal of Computer Vision,2004,0(2):91-110
[20] Van De Sande,Gevers K E A T,Snoek C G M.Evaluating colordescriptors for object and scene recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2010,32:1582-1596
[21] Hui Bin,Tang Xu-sheng,Luo Hai-bo,et al.SDF Matched Filter Based on Gabor Wavelet Transform for Face Recognition[J].Information and Control,2008,37(5):633-636

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!