Query based feature reduction and colour indexing scheme with fuzzy clustering for efficient CBIR in medical applications

Abstract

In this research the problem of content based retrieval has been analyzed and studied using various applications Modern information technology supports representation of the data to be performed with images As it provides stability to the content and supports the maintenance of the data to be performed in a rigid manner various domains and organizations use image based content management Content based retrieval has been used for indexing and retrieving of similar and exact images from a larger pool of image data bases The application of content based image retrieval has no limit as it has been applied in different applications like general image mining mining of government data base mining of criminal data set and medical applications The application of CBIR has no limit and can be adapted to any problem There are a number of methods available and many discussions have been made in literature regarding content based image retrieval The methods vary in the means measure and features being used However they fail to achieve the expected performance To improve the performance of CBIR this research has been carried out and it will contribute to design different algorithms First the dimensionality of the features of multimedia data set has been considered It is necessary to reduce the feature size so as to lessen time complexity with good retrieval accuracy Next a query based feature reduction base CBIR algorithm has been presented The method considers three different features namely: the texture feature which has been framed by the co occurrence matrix; the colour feature which has been represented by generating the dominant colour and the shape feature has been represented by region based features All the extracted features are normalized and reduced to minimum dimension by choosing only the specific features newline