摘要
提出了一种改进的模板匹配的数字识别算法,该算法是预先将字符分成若干个集合,经细化得到数字中央的骨骼部分,再对待识别数字提取特征并与训练库中的数字特征加权比较,利用欧式距离最小原则来对数字作出判决,试验结果表明,加权的模板匹配法保证了数字识别的正确率,而对数字进行预分类和细化处理,可以大大缩小模板匹配的识别速度,弥补了模板匹配算法对于大量数字耗时多的缺点,提升了系统速度。
An algorithm based on Improved Template matching method is proposed for digit recognition systems.The digit are divided into a number of small collection,then obtain the skeletal parts of the digit by the elaboration.The feature extraction of the digit is realized,and make weightea comparison with the features of the digits in the training library.Finally,the algorithm makes judgments according to the Minimum Euclidean distance principle.The results show that the template matching algorithm ensures the accuracy of digit recognition,the digit pre-classification and elaboration could reduce the digit searching range,save the time and improve the system speed.
出处
《工业控制计算机》
2010年第5期76-77,共2页
Industrial Control Computer
关键词
数字识别
特征量提取
模板匹配
digit recognition
feature extraction
template matching
