Image Categorization using Thepade's Sorted Block Truncation Coding N-ary with Data Mining Classifiers

Due to increase in use of social Networking sites uploading of multimedia data on internet has exponentially increased. The search and retrieval through such data can be made faster by categorizing the data according to predefined classes. Here the image classification technique is proposed using Thepade's Sorted Block Truncation Coding (TSTBTC) to classify the images into predefined classes. Image color content based signatures ares used for image classification. Assorted sizes of feature vectors are formed by taking Thepade's Sorted N-ary Block Truncation Coding. The various classification algorithms from assorted classifier families are deployed in proposed image classification techniques such as Function (Simple Logistic, RBF Network) Rule(Part, Decision), Tree (J48, BFTree, Random Forest, Random tree) and Bayes(BayesNet, Navie Bayes) the performance of various proposed techniques are analyzed using classification accuracy. Simple Logistic Classifier has given the highest classification accuracy of 87.5% with TSTBTC ‘Hexa ary’ version.