【摘要】 提出了一种用Bayes决策理论进行手写体数字识别的方法,对已知类别的样品提取特征建立数字样品库,对于任意的手写数字提取特征,根据数字样品库中已知样品的特征,运用基于最小错误概率的Bayes决策进行识别.实验证明Bayes决策理论用于手写体数字的识别有较好的效果,一般情况下识别率能达到96%以上.更多还原

【Abstract】 A new scheme for handwrittennumber recognition is proposed in this paper.According to features of samples whose classifications are known,a sample room is set up.To recognize a handwrittennumber,we should extract its features.Combiningwith the samples in sample room,its classification can be confirmed via Bayes decision based on the least error probability.Experiments show that the proposed scheme can classify handwritten number well,whose recognition rate is above 96%.更多还原