Abstract
Relevance feedback is commonly incorporated into content-based image retrieval systems with the objective of improving retrieval accuracy via user feedback. One effective method for improving retrieval performance is to perform feature re-weighting based on the obtained feedback. Previous approaches to feature re-weighting via relevance feedback assume the feature data for images can be represented in fixed-length vectors. However, many approaches are invalidated with the recent development of features that cannot be represented in fixed-length vectors. In addition, previous approaches use only the information from the set of images returned in the latest query result for feature re-weighting. In this paper, we propose a feature re-weighting approach that places no restriction on the representation of feature data and utilizes the aggregate set of images returned over the iterations of retrieval to obtain feature re-weighting information. The approach analyzes the feature distances calculated between the query image and the resulting set of images to approximate the feature distances for the entire set of images in the database. Two-sided confidence intervals are used with the distances to obtain the information for feature re-weighting. There is no restriction on how the distances are calculated for each feature. This provides freedom for how the feature representations are structured. The experimental results show the effectiveness of the proposed approach and in comparisons with other work, it is shown that our approach outperforms previous work.
Similar content being viewed by others
References
Aggarwal G, Ashwin TV, Ghosal S (2002) An image retrieval system with automatic query modification. IEEE Trans Multimedia 4–2:201–214
Cheikh FA, Cramariuc B, Gabbouj M (2003) Relevance feedback for shape query refinement. In Proceedings of the IEEE International Conference on Image Processing, September, Barcelona, Spain, pp 745–748
Ciocca G, Schettini R (2001) Content-based similarity retrieval of trademarks using relevance feedback. Pattern Recogn 34–8:103–119
Cox IJ, Miller ML, Minka TP, Papathomas TV, Yianilos PN (2000) The Bayesian image retrieval system, pichunter: theory, implementation, and psychophysical experiments. IEEE Trans Image Process 9–1:20–37
Faloutsos C, Flickner M, Hafner J, Niblack W, Petkovic D, Equitz W (1994) Efficient and effective query by image content. Journal of Intelligent Information Systems 3–3/4:133–150
Ghosh A, Petkov N (2005) Robustness of shape descriptors to incomplete contour representations. IEEE Trans Pattern Anal Mach Intell 27–11:1793–1804
Gondra I, Heisterkamp DR (2004) Learning in region-based image retrieval with generalized support vector machines. IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp 149–156
Hayter AJ (2002) Probability and statistics for engineers and scientists. Duxbury, Pacific Grove, CA
Heesch R, Ruger S (2003) Performance boosting with three mouse clicks—relevance feedback for CBIR. In 25th European Colloquium on IR Research, Springer LNCS 2633, pp 363–376
Huang P-W, Lee C-H (2004) Image database design based on 9D-SPA representation for spatial relations. IEEE Trans Knowl Data Eng 16–12:1486–1496
Ishikawa Y, Subramanya R, Faloutsos C (1998) Mindreader: querying databases through multiple examples. In Proceedings of the 24th International Conference on Very Large Data Bases, pp 218–227
Kim D-H, Chung C-W (2003) Qcluster: relevance feedback using adaptive clustering for content-based image retrieval. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pp 599–610
Kwon Y-I, Park H-H, Lee S-L, Chung C-W (2005) A shape feature extraction for complex topographical images. In Proceedings of the International Symposium on Remote Sensing, pp 575–578
Latecki LJ (2002) Shape data for the MPEG-7 Core Experiment CE-Shape-1. http://www.cis.temple.edu/~latecki/TestData/mpeg7shapeB.tar.gz
Lee K-M, Street WN (2004) Cluster-driven refinement for content-based digital image retrieval. IEEE Trans Multimedia 6–6:817–827
Lee RS, Kim S-H, Park H-H, Lee S-L, Chung C-W (2006) A feature re-weighting approach for the non-metric feature space. Journal of Korea Information Science Society: Databases 33–4:372–383
Manjunath BS, Ohm JR, Vasudevan VV, Yamada A (2001) Color and texture descriptors. IEEE Trans Circuits Syst Video Technol 11–6:703–715
Mokhtarian F, Bober M (2003) Curvature scale space representation: theory, applications, and MPEG-7 Standardization. Springer, New York
Porkaew K, Chakrabarti K, Mehrotra S (1999) Query refinement for multimedia similarity retrieval in MARS. Technical Report TR-DB-99-05, University of California at Irvine
Porkaew K, Mehrotra S, Ortega M (1999) Query reformulation for content based multimedia retrieval in MARS. Technical Report TR-DB-99-03, University of California at Irvine
Rui Y, Huang TS, Ortega M, Mehrotra S (1998) Relevance feedback: a power tool for interactive content-based image retrieval. IEEE Trans Circuits Syst Video Technol 8–5:644–655
Smith JR, Chang S-F (1997) VisualSEEk: A fully automated content-based image query system. In Proceedings of the ACM International Conference on Multimedia, pp 87–98
Taycher L, La Cascia M, Sclaroff S (1997) Image digestion and relevance feedback in the imagerover WWW Search Engine. In Proceedings of the 2nd International Conference on Visual Information Systems, pp 85–94
Tusk C, Koperski K, Aksoy S, Marchisio G (2003) Automated feature selection through relevance feedback. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, pp 3691–3693
Wu Y, Zhang A (2002) A feature re-weighting approach for relevance feedback in image retrieval. In Proceedings of the IEEE International Conference on Image Processing, pp 581–584
Acknowledgments
This research was supported by the Defense Acquisition Program Administration and the Agency for Defense Development, Korea, under the contract UD030000AD, through the Image Information Research Center at Korea Advanced Institute of Science and Technology.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Appendix 1 Proof of Theorem 1
Since the proof for Eq. 12 is straightforward, we will omit the proof for Eq. 12.
Every possible case for the placement of two confidence intervals in the negative range must be considered. Consider Fig. 10a where both bounds for interval 2 are greater than interval 1. Then shift the lower bound of interval 2 to be in the range of interval 1 to obtain case (b). Next, further shift the lower bound of interval 2 so that it is more negative than interval 1 to obtain case (c). All cases where the upper bound of interval 2 is greater than interval are covered. Thus, shift the upper bound of interval 2 into the range of interval 1. If the lower bound of interval 2 is also placed inside the range of interval 1, case (d) is obtained. Then, shift the lower bound of interval 2 so that it is more negative than interval 1 to obtain case (e). Finally, shift the upper bound of interval 2 to be more negative than interval 1, and this leaves only the possibility of the lower bound of interval 2 being more negative than interval 1 as displayed in case (f). As can be seen, case (c) and (d) are the same, as well as case (b) and (e), and likewise for cases (a) and (f). This leaves three cases that must be checked.
-
Case 1: Independent (Fig. 10a and f). Figure 11 illustrates the case where there are two confidence intervals that do not overlap.
Let
-
1.
lb1 + x = lb2,
-
2.
ub1 + y = ub2
where
-
3.
−1 < lb1 < ub1 ≤ 0,
-
4.
−1 < lb2 < ub2 ≤ 0,
-
5.
x > 0,
-
6.
y > 0.
Then
\(\operatorname{ub} _1 + ub_1 lb_1 + ub_1 x <\operatorname{ub} _2 + ub_2 lb_1 \) by substituting (1) \(\operatorname{ub} _1 + ub_1 lb_1 + ub_1 x <\operatorname{ub} _1 + y + \operatorname{ub} _1 lb_1 + lb_1 y\) by substituting (2) \(\operatorname{ub} _1 x <y + \operatorname{lb} _1 y\)
\(0 <y\left( {1 + lb_1 } \right) - \operatorname{ub} _1 x\) is true considering (3), (4), (5), (6).
-
Case 2: Overlapping (Fig. 10b and e)
Figure 12 illustrates the case where there are two confidence intervals that have partial overlap.
Let
-
1.
lb1 + x = lb2,
-
2.
ub1 + y = ub2
where
-
3.
−1 < lb1 < ub1 ≤ 0,
-
4.
−1 < lb2 < ub2 ≤ 0,
-
5.
x > 0,
-
6.
y > 0.
Starting from Eq. 16,
\(\operatorname{ub} _1 + ub_1 lb_1 + ub_1 x <ub_2 + ub_2 lb_1 \) by substituting (1)\(\operatorname{ub} _1 + ub_1 lb_1 + ub_1 x <ub_1 + y + ub_1 lb_1 y\) by substituting (2)
\(0 <y(1 + \operatorname{lb} _1 ) - \operatorname{ub} _1 x\) is true considering (3), (4), (5), (6).
-
Case 3: Contained (Fig. 10c and d)
Figure 13 illustrates the case where one confidence interval is fully contained in another confidence interval.
Let
-
1.
ub1 + x = ub2
where
-
2.
−1 < lb1 < ub1 ≤ 0,
-
3.
−1 < lb2 < ub2 ≤ 0,
-
4.
x > 0.
Starting from Eq. 16,
\(\operatorname{ub} _1 + ub_1 lb_2 <ub_1 + x + \operatorname{ub} _1 lb_1 + lb_1 x\) by substituting (1)\(0 <x\left( {1 + \operatorname{lb} _1 } \right)\) is true considering (2), (3), (4).
1.2 Appendix 2 MPEG-7 Core Experiment CE-Shape-1 Part B
The following is a subset of the images from the MPEG-7 Core Experiment CE-Shape-1.
Rights and permissions
About this article
Cite this article
Lee, R.S., Chung, CW., Lee, SL. et al. Confidence interval approach to feature re-weighting. Multimed Tools Appl 40, 385–407 (2008). https://doi.org/10.1007/s11042-008-0212-5
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11042-008-0212-5