Artificial Intelligent Systems and Machine Learning

K. Beningston ¹, K. Veningston ², J. Jacob Durai Raj ²

Affiliations
1 Department of Statistics, Bharathiar University, Coimbatore, IN
2 Department of Computer Science and Engineering, Karunya University, Coimbatore, IN

Abstract

Most of the image search engines nowadays use mainly text-based information. Since surrounding text is not always accurate, the returned images are often noisy and disorganized. Proposed work re-ranks current web search engine results according to user profile information. Content-based image retrieval (CBIR) uses visual feature to evaluate image similarity. However, due to the diversity of images and features, a universal feature set for all the images is hard to find. Relevance feedback uses user labeled images to improve image rank. However, most relevance feedback methods require online training based on feedback samples, and cannot be easily used for real-time online application. In this paper, we develop a technique to model user profile in order to regulate feature extraction and matching scheme to be used to re-organize images which has been retrieved by the web image search engine. The system has been designed to re-rank text-based search results in an interactive manner according to user intention.

Keywords

Image Search Engine, Content Based Image Retrieval, Relevance Feedback, Personalization, Image Re-Ranking.

Full Text