针对重轨生产线钢坯支支跟踪的需求,研究了一种基于计算机视觉的钢坯字符识别方法.该识别方法对在线采集到的钢坯字符图像采用基于最大类间方差的多级分割滤波与聚类处理突出字符目标区域,从而精准定位出钢坯字符;采用基于智能多代理者的切分算法来完成钢坯字符的精确切分;采用模板匹配与结构特征识别相结合的多级识别方法来正确识别出钢坯字符.实验结果表明所提出的算法能正确快速地识别出钢坯号字符.
To meet the demand of tracking each billet in the heavy rail production line,a billet character recognition algorithm based on computer vision was proposed.Firstly,multistage segmentation filtering based on OTSU and clustering processing was adopted to locate the billet character precisely.Secondly,the segmentation algorithm based on intelligent multi-agent was used to divide the billet character accurately.Lastly,the multilevel recognition algorithm incorporation the template matching and feature recognition was used to recognize billet character correctly.The experimental results show that the proposed algorithm in this paper can recognize the billet character correctly and quickly.