摘要
提出了基于方向梯度直方图(Histogramoforientedgradient,HOG)特征提取和模糊支持向量机(Fuzzysupportvectormachine,FSVM)的西夏文字识别技术.在模糊支持向量机模型中引入了新的隶属度函数,构造了基于多超平面的模糊支持向量机模型,增强了分类能力,降低了噪声点的干扰,提高了分类效率.将HOG特征提取和FSVM相结合应用于西夏文字识别,提高了文字识别效率.通过在数据集上测试,并与已有的文字识别方法相比较,结果表明,HOG特征提取结合FSVM的方法性能优于现有的其他方法.
A Tangut character recognition technology was proposed based on the feature extraction of histogram of oriented gradient(HOG) and fuzzy support vector machine(FSVM).In FSVM model,a new membership function was adopted to construct a fuzzy support vector machine based on multiple hyperplanes,enhances the classification ability,reduces the interference of noise points and improves the classification efficiency.The combination of HOG feature extraction and FSVM was applied to Tangut character recognition,which improved the efficiency of character recognition.Extensive experiments on various benchmarks show that the performance of HOG feature extraction combined with FSVM is better than other existing methods.
作者
刘兴长
孟昱煜
LIU Xing-chang;MENG Yu-yu(College of Electronic and Information Engineering,Lanzhou Jiaotong University,Lanzhou 730070,Gansu,China)
出处
《西北师范大学学报（自然科学版）》
CAS
北大核心
2019年第5期39-43,共5页
Journal of Northwest Normal University(Natural Science)
基金
国家自然科学基金资助项目(61662043)
甘肃省自然科学基金资助项目(1606RJZA033)
关键词
西夏文字识别
HOG特征提取
模糊支持向量机
Tangut character recognition
feature extraction
fuzzy support vector machine
