基于粗糙集的快速车牌字符识别技术

为了解决传统识别技术在车牌字符识别时效率低的问题,本文提出了一种基于粗糙集高效属性约简算法的快速车牌识别技术,该方法首先根据训练样本集的特征向量建立决策表并对决策表进行二次离散化处理,然后应用粗糙集理论对决策表进行高效属性约简,最后从约简后的决策表中获取决策规则,按照规则可信度的大小进行规则的匹配。实验表明该方法有效地压缩了图像的特征数,并简化了规则匹配算法,提高了字符识别率及识别速度,在车牌字符识别中取得了较好的识别效果。

A Method of Fast Recognizing Characters in Vehicle License Plates Based on the Theory of Rough Sets

In order to solve the problems of ineffectiveness about traditional method of fast recognizing characters in vehicle license-piates,A method for the recognition of vehicle license plate characters is presented based on efficient attribute reduction.The features of the training samples are extracted to build up the decision table and to execute tow-stage discretization.The rough sets theory is applied to reduce the decision att ributes of the table.Then the decision rules can be generated from the reduced decision table.The rule-match algorithm is based on the creditability of the rules.The application examples have showed a good result in the recognition of the vehicle＇s plate characters.The reduced decision attributes are beneficial to the improvement of the generality of the rules and the simplification of rule matching,to improve recognition rate and recognition speed.