Evaluation of best fit video visual classifier for Compacted Dither Pattern Codes

This paper presents the effectiveness of coupling of Compacted Dither Pattern Code (CDPC) visual syntactic feature with Bhattacharyya classifier for irregular shapes based visual concepts depiction. CDPC helps to reduce the feature space and complexity in classification. The results in this paper show that Bhattacharyya classifier able to further strengthen those capabilities. Here the Bhattacharyya classifier was compared with nu-SVM and k-NN classification methods which are well known visual classifiers among research community. The experimental results have revealed the capabilities of CDPC syntactic feature with the all three classifiers. The experiment was done for three irregular shape based visual concepts namely sky, grass and vegetation. The evaluation criteria and scheme was presented by using precision, recall and F1 score measures.