A text localization algorithm in images for automatic recognition of slab identification numbers

In the steel making process, each slab is given unique identification (ID) numbers for their management immediately after the continuous casting. Hundreds of slabs per day are pass through our target production lines and the ID numbers are identified by workers. The manual inspection of these ID numbers is a labor-intensive task and imposes heavy burden on the workers. In order to overcome these problems and to achieve a process automation, it is necessary to develop an automatic recognition system using optical character recognition technology. However, there are several difficulties in localizing text due to their variable position, variable color, and complex background in addition to considerable noise. For these difficulties, the overall performance of the recognition system is seriously affected by the performance of the text localization and extraction algorithms. In this paper, we propose Canny's edge-based text localization algorithm and presents the analysis of the experimental results.