Manuscript Number : IJSRST173619
An Enhanced Approach for Image Search Based on Attribute-Assisted Re-Ranking Model
Authors(2) :-Renuka S. Deshpande, Swati A. Patil Image search re-ranking is an effective approach to refine the text-based image search results. Most existing re-ranking approaches are based on low-level visual features. Here, semantic attributes are exploited for image search re-ranking. In the existing system, image features are extracted through K-means clustering and then the histogram is generated. The extracted features were then used for attribute learning using Support Vector Machine (SVM) classifier. After that the attribute assisted hyper-graph is constructed to re-rank the images. In the proposed system Keyword expansion using word Net and user keywords are used where word net gives the synonyms for the query and user keywords gives the similar meaning words for the improvement in the search result. Image features are extracted through discrete cosine transform and perceptual hashing. Then histogram is generated for all the images. Attribute learning is carried out using multi SVM. The new image re-ranking agenda focuses on the semantic signatures associated with the images derived using a trained multi class classifier. A query image is selected and then the most similar three images to the query image are displayed. Then these most similar images are further compared with the remaining images and are arranged in order according to their visual appearance and by comparing their hash values using Euclidean distance. The experiments are carried out on multiple data sets.
Renuka S. Deshpande Image Search, Image Re-Ranking, Semantic Signature, Keyword Expansion. Publication Details
Published in : Volume 3 | Issue 6 | July-August 2017 Article Preview
Computer Science and Engineering, GHRIEM Jalgaon Maharashtra, India
Swati A. Patil
Computer Science and Engineering, GHRIEM Jalgaon Maharashtra, India
Date of Publication : 2017-08-31
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Page(s) : 84-92
Manuscript Number : IJSRST173619
Publisher : Technoscience Academy
Journal URL : https://ijsrst.com/IJSRST173619
Citation Detection and Elimination |
|
| BibTeX | RIS | CSV