基于模板匹配的手写数字决策分类器设计  

Design of Handwritten Numeral Decision Classifier Based on Template Matching

作　　者：胡开华[1] 张玉静[1] 舒期梁[1] 

机构地区：[1]景德镇陶瓷学院科技艺术学院,景德镇333001

出　　处：《现代计算机》2011年第24期67-68,80,共3页Modern Computer

摘　　要：针对手写数字识别提出一种基于模板匹配决策分类器设计方法。就该方法下的模式识别分类器设计进行详细论述,给出该分类器算法实现。该算法在对手写的数字图像进行预处理的基础上从待识别的手写数字图像中提取若干特征量与事先建立的标准模板库中模板对应的特征量进行比较,计算待识别图像和标准模板特征量之间的距离,用最小距离法判定其所属类。实验结果表明,该决策分类器算法实现容易,匹配速度快,保证字符识别的正确率。Gives a design method based on template matching decision classifier for handwritten numeral recognition. Discusses the methods of pattern recognition classifier design, gives the classifier algorithm. The algorithm based on handwritten digital image pre processing extracts some characteristic quantities from the recognition of handwritten digital image and compares with standard template library of templates corresponding to the feature previously established. Calculates the distance between identifying image and the standard template features. Uses minimum distance method to determine the class it belongs to. The experimental result shows that, the decision classifier algorithm is easy to realize, and has the matching speed, so as to ensure the correct rate of the character recognition.

关 键 词：模板匹配 分类决策 手写数字识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]