摘要
该文针对皮肤显微图像症状识别过程中样本采集困难、数量偏少的实际情况,在皮肤症状识别中引入了一种新的模式识别方法———支持向量机(SupportVectorMachines,SVM).该方法基于统计学习理论的原理,较好地解决了小样本的分类问题.文中采用"一对一"的策略解决多类别的SVM分类问题,使用留一法进行交叉验证,并比较了SVM与人工神经网络算法的识别结果.结果表明,SVM算法识别率高(89.35%),且速度快.根据该算法,建立了皮肤症状显微图像识别系统软件的原型.
Traditional recognition methods for microscopic images have satisfactory classification performance only when the number of samples is large enough. However this requirement usually cannot be met. In this paper, a new recognition method based on support vector machines (SVM) is proposed to solve the learning problem with a small sample size. Furthermore, a leave-one-out cross-validation (LOOCV) scheme is employed to test the performance of SVM classification, and the 'one-against-one' approach is used to solve multi-class classification problems. Using this new method, cross validation accuracy of 89.35% has been obtained in the classification of microscopic images of human skin, suggesting that the SVM approach is better than that of neural networks.
出处
《上海大学学报(自然科学版)》
CAS
CSCD
北大核心
2005年第1期24-27,共4页
Journal of Shanghai University:Natural Science Edition
基金
上海市科委基础研究项目 (0 2DJ1 40 3 4)
上海市科委技术攻关项目 (0 2 591 1 3 2 3 )
关键词
支持向量机
皮肤显微图像
模式识别
分类
support vector machines
microscopic image of skin
pattern recognition
classification