SVM Employed Accurate and Efficient Fusion Based Image Retrieval Model

Shefali Dhingra

Abstract: In this paper, an accurate image retrieval system is proposed based on the fusion of features. The complete system is framed by the combination of color and texture techniques i. e. color moment and Local binary pattern respectively. To make the framed system more efficient Support vector machine is applied as a classifier on the fused feature vector. The system is named as CT - SVM. These experiments are performed on benchmark dataset i. e. Corel - 1K. The results are calculated for the proposed system and the obtained results for CT - SVM system shows an appreciable improvement in terms of average precision that is 90% in case of Corel - 1K dataset.

Keywords: Color moment CM local binary pattern LBP, Support vector machine SVM