NOVATEUR PUBLICATIONS

International Journal Of Research Publications In Engineering And Technology [IJRPET]

ISSN: 2454-7875
VOLUME 3, ISSUE 2, Feb. -2017

MINING USER NAVIGATION PATTERNS FOR EFFICIENT RELEVANCE

FEEDBACK FOR CBIR
MR. SHINDE SURESH GOROBA
M.E. Scholar, Department of Computer Science & Engineering, Bharat Ratna Indira Gandhi College of Engineering, Solapur

PROF. C. M. JADHAV
Assistant Professor, Department of Computer Science & Engineering, Bharat Ratna Indira Gandhi College of Engineering,
Solapur

ABSTRACT: Typically, in the development of an image requisition

In today’s modernized world, content based system, semantic image retrieval relies heavily on the
image retrieval (CBIR) is considered as a bastion in related captions, e.g., categories, file-names, and other
image retrieval system. For making CBIR most manual descriptions, annotated keywords.
suitable and productive technique, relevance Unfortunately, this kind of textual-based image retrieval
feedback technique is used in conjunction with CBIR suffers from two problems: high-priced manual
for producing more specific results which are annotation and inappropriate automated annotation.
obtained by taking feedback from user. However,
existing relevance feedback-based CBIR methods PROPOSED SYSTEM:
usually request a number of iterative feedbacks for The proposed algorithm NPRF Search performs
production of best search results, particularly in the navigation-pattern-based search to match the user’s
huge database. But this seems of no use in real world intention by merging three query refinement strategies.
applications. In this paper, we propose a novel As a result, traditional problems such as visual diversity
method, NRPF (Navigation pattern based relevance and exploration convergence are solved. For navigation-
feedback method is used for enhancing effectiveness pattern-based search, the hierarchical BFS based KNN is
and efficiency in CBIR while copying large scale employed to narrow the gap between visual features and
image data. In terms of efficiency, the iterations of human concepts effectively. In addition, the involved
feedback will get reduced drastically reduced methods for special data partition and pattern pruning
substantially by using the navigation patterns also speed up the image exploration. The experimental
discovered from the user query log. Effectiveness of results reveal that the proposed approach NPRF is very
our proposed search algorithm NPRF Search makes effective in terms of precision and coverage. With short
use of the discovered navigation patterns and it also term relevance feedback, the navigation system
produces query refinement strategies in other three algorithm will help in assisting the users in obtaining the
kinds, Query Point Movement (QPM), Query best results. Moreover, the new search algorithm NPRF
Expansion (QEX) and Query Reweighting (QR), to Search can bring out more accurate results than other
converge the search space toward the user’s well-known approaches.
intention effectively. For this purpose NPRF systems
are used for increasing quality of retrieved image.
The experimental shows NPRF outperforms other
established methods considerably in terms of
precision, coverage, and number of feedbacks.
KEYWORDS: Content-based image retrieval,
relevance feedback, query point movement, query
expansion, navigation pattern mining etc.

INTRODUCTION:
Multimedia contents are growing explosively
and the need for multimedia retrieval is occurring more Fig. No.1. Example of navigation pattern trees [18]
and more frequently in our daily life. Understanding the
image has become difficult but that has raised the WORKING OF THE NRPF METHOD:
interest in this domain. Extracting valuable knowledge INITIAL QUERY PROCESSING PHASE:
from a large-scale multimedia repository, so-called Without considering the feature weight, this
multimedia mining, has been studied by few researchers. phase extracts the visual features from the original
4|P a g e
 NOVATEUR PUBLICATIONS
International Journal Of Research Publications In Engineering And Technology [IJRPET]
ISSN: 2454-7875
VOLUME 3, ISSUE 2, Feb. -2017
query image to find the similar images. Afterward, the Note that the procedure of constructing rule base from
good examples (also called positive examples in this the image databases can be conducted periodically to
paper) picked up by the user are further analyzed at the maintain the validity of the proposed approach.
first feedback (also called iteration 0 in this paper).
SYSTEM REQUIREMENTS:
IMAGE SEARCH PHASE: HARDWARE REQUIREMENTS:
Behind the search phase, our intent is to extend  System : Pentium IV 2.4
the one search point to multiple search points by GHz.
integrating the navigation patterns and the proposed  Hard Disk : 40 GB.
search algorithm NPRF Search. Thus, the varied  Monitor : 15 VGA Color.
inclusion of the user’s interest can be successfully  Mouse : Logitech.
implied. In this phase, a new query point at each  Ram : 512 MB.
feedback is generated by the preceding positive
examples. Then, the k-nearest images to the new query SOFTWARE REQUIREMENTS:
point can be found by expanding the weighted query.  Operating System : Windows xp ,
The search procedure does not stop unless the user is Linux
satisfied with the retrieval results  Language : Java1.4 or more
 Technology : Swing, AWT

CONLCUSION:
The major difference between our proposed
approach and other contemporary approaches is that we
approximate an optimal solution to resolve the problems
existing in current RF, such as redundant browsing and
exploration convergence. To this end, the approximated
solution takes advantage of exploited knowledge
(navigation patterns) to assist the proposed search
strategy in efficiently hunting the desired images.

REFERENCES:
1) M.D. Flickner, H. Sawhney, W. Niblack, J. Ashley, Q.
Huang, B.Dom, M. Gorkani, J.Hafner, D. Lee, D.
Steele, and P. Yanker, “Query by Image and Video
Content: The QBIC System,” Computer, vol. 28, no. 9,
Fig.no.1 Work flow of NRPF method pp. 23-32, Sept. 1995.
2) R. Fagin, “Combining Fuzzy Information from
KNOWLEDGE DISCOVERY PHASE: Multiple Systems,” Proc. Symp. Principles of
Learning from users’ behaviors in image Database Systems (PODS), pp. 216-226, June 1996.
retrieval can be viewed as one type of knowledge 3) R. Fagin, “Fuzzy Queries in Multimedia Database
discovery. Consequently, this phase primarily concerns Systems,” Proc. Symp. Principles of Database
the construction of the navigation model by discovering Systems (PODS), pp. 1-10, June 1998.
the implicit navigation patterns from users’ browsing 4) J. French and X-Y. Jin, “An Empirical Investigation of
behaviors. This navigation model can provide image the Scalability of a Multiple Viewpoint CBIR System,”
search with a good support to predict optimal image Proc. Int’l Conf. Image and Video Retrieval (CIVR),
browsing paths. pp. 252-260, July 2004.Cheng, L. and Adams, D.L.,
1995. Yarn strength prediction using neural
DATA STORAGE PHASE: networks, Text. Res. J., 65(9), 495-500.
The databases in this phase can be regarded as 5) D. Harman, “Relevance Feedback Revisited,” Proc.
the knowledge marts of a knowledge warehouse, which 15th Ann. Int’l ACM SIGIR Conf. Research and
store integrated, time-variant, and nonvolatile collection Development in Information Retrieval, pp. 1-10,
of useful data including images, navigation patterns, log 1992.
files, and image features. The knowledge warehouse is 6) Y. Ishikawa, R. Subramanya, and C. Faloutsos,
very helpful to improve the quality of image retrieval. “Mind Reader: Querying Databases through Multiple
5|P a g e
 NOVATEUR PUBLICATIONS
International Journal Of Research Publications In Engineering And Technology [IJRPET]
ISSN: 2454-7875
VOLUME 3, ISSUE 2, Feb. -2017
Examples,” Proc. 24th Int’l Conf. Very Large Data Experiments in Automatic Document Processing,
Bases (VLDB), pp. 218-227,1998. pp. 313-323, Prentice Hall, 1971.
7) X. Jin and J.C. French, “Improving Image Retrieval 14) Y. Rui, T. Huang, and S.Mehrotra, “Content-Based
Effectiveness via Multiple Queries,” Multimedia Tools Image Retrieval with Relevance Feedback in MARS,”
and Applications, vol. 26, pp. 221-245, June 2005. Proc. IEEE Int’l Conf. Image Processing, pp. 815-
8) D.H. Kim and C.W. Chung, “Qcluster: Relevance 818, Oct. 1997.
Feedback Using Adaptive Clustering for Content- 15) Y. Rui, T. Huang, M. Ortega, and S. Mehrotra,
Based Image Retrieval,” Proc. ACM SIGMOD, pp. “Relevance Feedback: A Power Tool for Interactive
599-610, 2003. Content-Based Image Retrieval,” IEEE Trans.
9) K. Porkaew, K. Chakrabarti, and S. Mehrotra, “Query Circuits and Systems for Video Technology, vol. 8,
Refinement for Multimedia Similarity Retrieval in no. 5, pp. 644-655, Sept. 1998.
MARS,” Proc. ACM Int’l Multimedia Conf. (ACMMM), 16) P.Y. Yin, B. Bhanu, K.C. Chang, and A. Dong,
pp. 235-238, 1999. “Integrating Relevance Feedback Techniques for
10) J. Liu, Z. Li, M. Li, H. Lu, and S. Ma, “Human Behavior Image Retrieval Using Reinforcement Learning,”
Consistent Relevance Feedback Model for Image IEEE Trans. Pattern Analysis and Machine
Retrieval,” Proc. 15th Int’l Conf. Multimedia, pp. Intelligence, vol. 27, no. 10, pp. 1536-1551, Oct.
269-272, Sept. 2007. 2005.
11) A. Pentalnd, R.W. Picard, and S.Sclaroff, 17) V.S. Tseng, J.H. Su, J.H. Huang, and C.J. Chen,
“Photobook: Content- Based Manipulation of Image “Integrated Mining of Visual Features, Speech
Databases,” Int’l J. Computer Vision (IJCV), vol. 18, Features and Frequent Patterns for Semantic Video
no. 3, pp. 233-254, June 1996. Annotation,” IEEE Trans. Multimedia, vol. 10, no. 2,
12) T. Qin, X.D. Zhang, T.Y. Liu, D.S. Wang, W.Y. Ma, and pp. 260-267, Feb. 2008
H.J. Zhang, “An Active Feedback Framework for 18) Satyanarayana Mummana , Swathi Koundinya ,
Image Retrieval,” Pattern Recognition Letters, vol. “Navigation-Pattern-Based Relevance Feedback For
29, pp. 637-646, Apr. 2008. Content Based Image Retrieval” International
13) J.J. Rocchio, “Relevance Feedback in Information Journal Of Engineering Science & Advanced
Retrieval,” The SMART Retrieval System— Technology. Volume-2, Issue-5, 1241 – 1246

6|P a g e