Mark identification of proofed steel products

Abstract

A computer vision based recognition system is proposed to identify the mark of proofed steel product in this study. The identification process uses image processing with a template matching to locate and recognize marking letters and numbers. The two-dimensional Haar discrete wavelet transform (DWT) is used to achieve image pyramids, so that computational efficiency is improved significantly. The proposed auto-identification procedure can effectively adapted to different steel product according to a specified template.