Exploiting Pre-calculated Distances in Nearest Neighbor Search on Query Images for Cbir

Nearest neighbor algorithms can be implemented on content-based image retrieval (CBIR) and classification problems for extracting similar images. In k-nearest neighbor (k-NN), the winning class is based on the k nearest neighbors determined by comparing the query image against all training samples. In this paper, a new nearest neighbor search (NNS) algorithm is proposed using a two-step process. The first step is to calculate the distances between all training samples. The second step is to discriminate samples falling outside the potential region of being a nearest neighbor, hence reducing the computation required. Experimental results showed that the proposed algorithm is able to obtain all nearest neighbors within the defined search radius. Therefore, the classification rate is identical to k-NN but less training samples are compared. It is shown that only 27.13% of the training samples are computed for 1024 Brodatz textures of thirty-two classes at a radius of 0.2 and 0.56% for a single best neighbor search. Experimental results also showed that the proposed algorithm is faster than k-NN during a single best neighbor search but significantly slower for a search with a defined search radius.