CONTENT-BASED IMAGE RETRIEVAL BASED ON SPELLING IMAGES

Abstract

Nowadays, image retrieval have a new age, content-based image retrieval age. Resolving the problem of content-based image retrieval is the challenge must been overcome in our life. The key problem in achieving an efficient and user-friendly system is the development the description of query data close to human beings and athering the minimal irrelevant information together the relevant information is not overlooked. We have solved this problem by discovering the implicit high-level features in image database. These features are extracted by segmenting the image into the regions and clustering them into the clusters with their representatives that we called the words of image. We have used the Hierarchical Agglomerative Clustering algorithm to segment the images, cluster the regions amd cluster the images. The query data can be the whole image, or some regions of query images or some region cluster's representatives. Retrieving is performed on the region clusters's hierarchical tree structure or the image cluster's hierarchical tree structure. Our experiment results have shown that our retrieval system is efficient and laser-friendly.