Design and Implementation of the Nearest Neighbour Search Algorithm for Image Searching Using Vector Quantization and Clustering Approach

ABSTRACT

In this paper, we analyse a high-dimensional space of linked characteristics representing picture datasets, and we investigate methodologies based on clustering and vector quantization for identifying precise similarity between them. Using hyperplane splitting, our index is able to return only clusters containing data items that are close to the query, which results in a reduction in the total number of results that are returned. As long as the same number of sequential disc accesses is permitted and pre-treatment storage and computational charges are kept to a bare minimum, proposed solution is more efficient than the conventional Vector Approximation File (VA-File) approach. The proposed approach is the Nearest Neighbour search algorithm designed for image Searching.

FULL TEXT