Content based Image Retrieval using the Domain Knowledge Acquisition

It is widely known that, nowadays the whole world is dominated by the visual information and this keeps increasing day by day. These data should be organized in such a way that the image retrieval can be more effective as well as efficient, this can be carried out through CBIR method. This paper has proposed a novel methodology named as DKA (Domain Knowledge Acquisition), which provides the advanced image retrieval as it considers the manifold structure. Moreover, the other existing technique requires the complex segmentation region from background. This research work are carried as first the shape feature has been used and once it is done a SIFT feature is applied and later with the help of DKA(Domain Knowledge Acquisition) the features are redefined and finally it helps in capturing the manifold structure in a better way and expose the inherent relation among the objects. The main intention here is to develop the efficient CBIR technique. In order to evaluate the algorithm, the feature point analysis has been performed with the help of Corel-10k datasets and later the comparison is done with the default SIFT feature and the SIFT with DKA. The analysis shows that, SIFT with DKA provides more flexible and efficient technique.