Abstract
A new efficient dynamic programming (DP) algorithm for 2D elastic matching is proposed. The present DP algorithm requires by far less complexity than previous DP-based elastic matching algorithms. This complexity reduction results from piecewise linearization of a 2D-2D mapping, which specifies elastic matching between two given images. Since this linearization can be guided by a priori knowledge related to image patterns to be matched, the present DP algorithm often provides sufficient matching as is shown by experimental results.