Content Based Image Retrieval using Shape Based Region Moments with Support Vector Machine Classifier

Abstract

Content Based Image Retrieval (CBIR) using shape based region moments with a support vector machine is successfully implemented in this paper. The recognition is purely based on the three different models such as color, shape and texture. One of the powerful methods is shape based feature extraction. Complex boundaries are effectively represented by moments. Complex boundaries depend  on the contour pixels and all the pixels that make up the shape. Due to the properties of orthogonal invariance   and geometric moment, the features are extracted from the huge training dataset using various moments. CBIR has been implemented on the COREL database and the SVM classifier. The performance measure is recognition efficiency, which ios the best metric to assess the  performance.