Faster neighbor finding on images represented by bincodes

The Interpolation-Based Bintree (IBB) is a new encoding scheme for representing binary images and is storage-saving. In the IBB only the black leaves are saved, called bincodes, as the compressed codes for the image. In this paper a fast neighbor finding algorithm is presented on bincodes. Given a sequence of bincodes with size n, our algorithm is performed in O((l(max) - l(min) + 1) x n) time with the same memory complexity, where l(max) and l(min) are the maximal level and the minimal level, respectively, at which the given bincodes reside. Our algorithm is faster than the previous one [C.-Y. Huang and K.-L. Chung, Pattern Recognition 28(3), 409-420 (1995)]. Experimental results for a practical version are included. The experimental values confirm our theoretical results. Copyright (C) 1996 Pattern Recognition Society. Published by Elsevier Science Ltd.