摘要
模式识别诞生于20世纪20年代。随着40年代计算机的出现、50年代人工智能的兴起,模式识别在60年代迅速发展成一门学科。模式识别系统主要由4个部分组成:数据采集及预处理,特征生成,特征选择与提取,分类器的设计与识别。主要应用方法分为两类:几何(非参数法)和概率(参数法)。其中几何方法包括:最小距离法、近邻法、梯度下降法、神经网络以及支持向量机。概率法包括:最小错误率Bayes决策,最小风险Bayes决策、聂曼-皮尔逊法。本文运用最小距离法和近邻法对数字图像进行识别,进而对两种方法做出比较。
Pattern Recognition was born in the 1920s.In the 60s,it quickly developed into a discipline with the emergence of computer in the 40s and the rise of artificial intelligence.Pattern recognition system consists of four parts： data acquisition and preprocessing,feature generation,feature selection and extraction,classifier design and identification.The main application method is divided into two categories： geometric（non-parametric method） and probability（parameter method）.Geometric approach includes： minimum distance method,nearest neighbor method,gradient descent method,neural networks and support vector machines.Probability method includes：minimum error rate Bayes decision,the minimum risk Bayes decision-making,Neyman-Pearson method.The paper uses the minimum distance method to recognize digital image.
出处
《仪器仪表用户》
2012年第5期53-55,共3页
Instrumentation
