一种新的激光刻蚀字符的识别方法

通过对铭牌字符结构特征和字符之间相对关系的分析,提出基于神经网络和隐马尔可夫模型的激光刻蚀字符的识别方法。提取字符图像中的端点、三叉点和四叉点,对传统三叉点的提取方法进行改进,并利用神经网络的方法计算出对应每个字符的概率值,再根据隐马尔可夫模型计算出状态转移的最大似然,从而识别整个字符串。实验表明上述方法适用于激光刻蚀铭牌字符的识别。

A New Recognition Method of Laser Etched Characters

Through the analysis of the relationships between structural properties and the nameplate characters,a new recognition method of laser etched characters is proposed based on Hidden Markov models and Neural Network.The traditional method based on three crossing points extraction is improved by taking into account the endings,three crossing points,four crossing points of the analyzed characters.Neural networks are employed to estimate probabilities for the characters depending on their structural properties,and Markov models are used to derive the best word choice from a sequence of state transition.It is shown in test that the proposed method can be used to recognize the etched characters on metal label.