Relevance Feedback is an important tool for grasping user's need in Interactive Content Based Image Retrieval (CBIR). Keeping this in mind, we have build up a framework using Relevance Vector Machine Classifier in interactive framework where user labels images as appropriate and inappropriate. The refinement of the images shown to the user is done using a few rounds of relevance feedback. This appropriate and inappropriate set then provides the training set for the RVM for each of these rounds. The method uses Histogram Intersection kernel with this interactive RVM (IKRVM). It has a retrieval component on top of this which searches for those images for retrieving which falls in the nearest neighbor set of the query image on the basis of histogram intersection based identical ranking (HIIR). The experimental results shows that the proposed framework shows better precision when compared with Active learning based RVMActive implemented with Radial Basis or Polynomial Kernels.
VAISHNAVI, K. and KUMAR, G.P.RAMESH
"CONTENT BASED IMAGE RETRIEVAL ASSOCIATE WITH THE INTERSECTION OF MULTI KERNAL RELEVANCE VECTOR MACHINE AND HISTOGRAM JOIN,"
International Journal of Computer and Communication Technology: Vol. 6:
4, Article 4.
Available at: https://www.interscience.in/ijcct/vol6/iss4/4