Road Sign Recognition Algorithm using Levenshtein Distance with the Least Squares Method

Abstract

An original road sign recognition algorithm using Levenshtein distance was introduced in 2007 by the authors. A new recognition algorithm combined Levenshtein distance with the least squares method enables to improve the recognition ratio is proposed. While the original algorithm could not detect a road sign hidden by trees, the new algorithm based on a shape judgment can find the road sign more reliable. The recognition ratio has been improved by 20% compared with the original algorithm. The new algorithm is implemented on Xilinx Spartan-3 using Bach-C at 49.8 MHz and achieved 17.0 ms execution time.