Cluster Oriented Image Retrieval System with Context Based Color Feature Subspace Selection

Barakbah, Ali Ridho (2010) Cluster Oriented Image Retrieval System with Context Based Color Feature Subspace Selection. Industrial Electronic Seminar.

Preview

PDF - Published Version Download (852Kb) | Preview

Abstract

This paper presents a cluster oriented image retrieval system with context recognition mechanism for selection subspaces of color features. Our idea to implement a context in the image retrieval system is how to recognize the most important features in the image search by connecting the user impression to the query. We apply a context recognition with Mathematical Model of Meaning (MMM) and then make a projection to the color features with a color impression metric. After a user gives a context, the MMM retrieves the highest correlated words to the context. These representative words are projected to the color impression metric to obtain the most significant colors for subspace feature selection. After applying subspace selection, the system then clusters the image database using Pillar-Kmeans algorithm. The centroids of clustering results are used for calculating the similarity measurements to the image query. We perform our proposed system for experimental purpose with the Ukiyo-e image datasets from Tokyo Metropolitan Library for representing the Japanese cultural image collections.

Actions (login required)