Waleed Farag
Content-based Image Retrieval, Evolutionary Algorithms, Optimizing Feature Space, Color features, Genetic Operators
The pervasive generation and use of various Multimedia contents emphasize the fact that these systems are now integrated parts of our life. As a result, organizing and accessing such data have been the focus of many researchers in our field over the past years. Although many proposed systems have been successful, the need for contemporary methodologies and tools that can cope with advances in media sizes and representations is still a necessity. This paper introduces a new system for Content-Based Retrieval (CBR) of image data. The proposed system presents a novel use of the genetic algorithms in order to improve a number of performance assessment criteria. The evolutionary algorithm optimizes a set of given features in the feature space and ensures that only promising candidates are returned as matches to the submitted query. Our hypothesis is that the integration of the GAs with an image CBR system will not only enhance the effectiveness of the system in returning best matches and avoiding irrelevant ones but also will improve the efficiency of the system by excluding non-promising areas of the search space which yields faster response. The Experimental results confirmed our hypothesis and showed promising performance in terms of convergence to best matches, efficacy and retrieval accuracy and precision.
Important Links:
Go Back
