Content Based Image Recognition by Information Fusion with Multiview Features

Substantial research interest has been observed\udin the field of object recognition as a vital component for\udmodern intelligent systems. Content based image\udclassification and retrieval have been considered as two\udpopular techniques for identifying the object of interest.\udFeature extraction has played the pivotal role towards\udsuccessful implementation of the aforesaid techniques.\udThe paper has presented two novel techniques of feature\udextraction from diverse image categories both in spatial\uddomain and in frequency domain. The multi view features\udfrom the image categories were evaluated for\udclassification and retrieval performances by means of a\udfusion based recognition architecture. The\udexperimentation was carried out with four different\udpopular public datasets. The proposed fusion framework\udhas exhibited an average increase of 24.71% and 20.78%\udin precision rates for classification and retrieval\udrespectively, when compared to state-of-the art\udtechniques. The experimental findings were validated\udwith a paired t test for statistical significance