CAJ | 학술논문

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剐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。
침대수사수자식별제출일충기우모판필배결책분류기설계방법。취해방법하적모식식별분류기설계진행상세논술，급출해분류기산법실현。해산법재대수사적수자도상진행예처리적기출상종대식과적수사수자도상중제취약간특정량여사선건립적표준모판고중모판대응적특정량진행비교，계산대식별도상화표준모판특정량지간적거리，용최소거리법판정기소속류。실험결과표명，해결책분류기산법실현용역，필배속도쾌，보증자부식별적정학솔。
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.