Content Bases Image Search And Retrieval Using Indexing By KMeans Clustering Technique

Abstract: With the popularity of network and development of multimedia technology, the traditional information retrieval techniques are not working efficiently according to users demand in searching and retrieving images from database. Recently the content based image retrieval concept becomes the hot topic in information retrieval domain. Due to the demand of information retrieval technique for image retrieval research has focused on content based image retrieval method. In today’s world there is increased need of content based image retrieval technique in number of different domains such as education, medical imaging, crime prevention, whether forecasting, remote sensing and management of earth resources. Content based image retrieval (CBIR) deals with retrieval of relevant images from the large image database. It works on the features of images like color and texture. In our system we are proposing an enhancement to basic content based image retrieval technique with indexing support by using K-means clustering data mining technique. The enhanced feature helps in retrieving images from large database fastly. In this system an index is applied on database of images based on clustering technique. During this process clustering concept uses features like texture, color, shape, relevance feedback and wavelet based histogram method to find similarity among the images. Based on similarity value the images are divided into clusters, then the new image which is to be verified with database is compared with these clusters and based on its similarity corresponding images in cluster are retrieved.

Keywords: Image retrieval, clustering, color, texture, histogram, similarity matching, semantic similarity, K-means.