A novel optimized image feature selection algorithm using pairwise classifiers
Mahdi Bazarganigilani, Ben Arko, Ali Asghar Mehdi Syed, Camille C. Price, Sandid Burki, Belinda Fridey, Michael Baron · Charles Sturt University Research Output (CRO) · 2013
This paper introduces an optimized method to improve the accuracy of content based image retrieval systems (CBIR). CBIR systems classify the images according to low and higher features. This research improves both feature selection and classifier partition of a CBIR system. This paper normalizes and descretizes the feature vectors. A pair-wise classifier is employed to improve the accuracy of classification. Results show high accuracy of proposed classifier in an image database.