In content-based image retrieval systems, image contents are represented as feature vectors and a distance between them is used as a similarity measure bewteen corresponding images. Thus, searching similar images is analogous to the k-nearest-neighbor...
In content-based image retrieval systems, image contents are represented as feature vectors and a distance between them is used as a similarity measure bewteen corresponding images. Thus, searching similar images is analogous to the k-nearest-neighbor searching problem. For this problem low-dimensional transformation techniques were proposed, which are used when dimensionality of the feature vector is very high or the distance measure is not so simple as a Euclidean distance. These techniques map high dimensional vectors into a lower dimensional space and evaluate the original distance metric only for those vectors within some bound in the lower dimension. This reduces the number of the high dimensional matching, However, to guarantee no false dismissals the existing methods use a maximum search space, which may be too wider than is practically needed.
We propose an incremental filtering algorithm that can reduce the search space. The proposed algorithm first chooses a minimal search space and incrementally expands it until the target number of results is obtained. This algorithm also guarantees no false dismissals and lower-bounds the existing methods with respect to the number of the evaluations of the expensive measure.