23-05-2012, 11:00 AM
IMAGE RETRIEVAL USING MEMORY LEARNING FRAMEWORK
INTRODUCTION
Content Based Image Retrieval
DUE to the rapidly growing amount of digital image data on the Internet and in digital libraries, there is a great need for large image database management and effective image retrieval tools. Content-based image retrieval (CBIR) is the set of tech¬niques for searching for similar images from an image database using automatically extracted image features.
Tremendous research has been devoted to CBIR and a variety of solutions have been proposed within the past ten years. By and large, research activities in CBIR have progressed in three major directions: global features based, object/region-level features based, and relevance feedback. Initially, developed sys¬tems are usually based on the carefully selected global image features, such as color, texture or shapes, and prefixed similarity measure. They are easy to implement and perform well for images that are either simple or contain few semantic contents (for example, medical images and face images). How¬ever, for these systems, it is impossible to search for objects or regions of the image. Therefore, the second group of sys¬tems is proposed on image segmentation. Contrasting
The performance of these systems mainly relies on the results of segmentation. Therefore, they cannot generate ex¬tremely good performance since the image segmentation is still an open problem in computer vision so far.
The limited retrieval accuracy of image-centric retrieval sys¬tems is essentially due to the inherent gap between semantic concepts and low-level features. In order to reduce the gap, the interactive relevance feedback (RF) is introduced into CBIR. RF, originally developed for textural document retrieval, is a supervised learning algorithm used to improve the performance of information systems. Its basic idea is to incorporate human perception subjectivity into the query process and provide users with the opportunity to evaluate the retrieval results. The sim-ilarity measures are automatically refined on the basis of these evaluations. After RF for CBIR was first proposed by Rui et al., this area of research has attracted much attention and become active in the CBIR community.
Recently, many researchers began to consider the RF as a learning or classification problem. That is, a user provides posi¬tive and/or negative examples, and the systems learn from such examples to refine the retrieval results or train a classifier by the labeled examples to separate all data into relevant and irrelevant groups.
Although RF can significantly improve the retrieval perfor¬mance, its applicability still suffers from three inherent draw¬backs.
1) Incapability of capturing semantics.
2) Scarcity and imbalance of feedback examples.
3) Lack of the memory mechanism.
Existing Method
• Most current content-based image retrieval systems are still incapable of providing users with their desired results.
• The major difficulty lies in the gap between low-level image features and high-level image semantics. To address the problem, this study reports a framework for effective image retrieval by employing a novel idea of memory learning.
• It forms a knowledge memory model to store the semantic information by simply accumulating user-provided interactions.
• Incapability of capturing semantics. Most RF tech¬niques in CBIR absolutely copy ideas from textural information retrieval.
• Scarcity and imbalance of feedback examples.
• Very few users are willing to go through endless iterations of feedback with the hopes of getting the best results. Lack of the memory mechanism.
Proposed System
A feedback knowledge memory model is presented to gather the users' feedback information during the process of image search and feedback. It is efficient and can be simply implemented.
A learning strategy based on the memorized information is proposed. It can estimate the hidden semantic relation¬ ships among images. Consequently, this technique could address the problem of user log sparsity in a certain ex¬tent.
During the interactive process, a seamless combination of normal RF (low-level feature based) and the memory learning (semantics based) is proposed to improve the
Retrieval performance. Notice that this combination is not a pure linear summation. The memory learning provides the normal RF with a pool of positive examples according to its captured knowledge, which helps the normal RF to alleviate the problem of scarcity and imbalance of feed¬
back examples.
A semantics-based image annotation propagation scheme is proposed using both memorized and learned semantics. In contrast with existing algorithms of propagating anno¬tation by visual similarity, its precision is much better. The rest of this paper is organized as follows. We briefly review the related work. In Section III, we present the feedback knowledge memory model. The learning strategy to estimate the hidden semantics is described. The image retrieval framework by memory learning is explained.
SYSTEM SPECIFICATION:
HARDWARE SPECIFICATION:
Processor : Intel Pentium-IV
Speed : 1.1GHz
RAM : 512MB
Hard Disk : 40GB
General : Key Board, Monitor , Mouse
SOFTWARE SPECIFICATION:
Operating System : Windows XP
Software : JAVA ( JDK 1.5.0)