摘要
支持向量机是基于统计学习理论的模式分类器。它通过结构风险最小化准则和核函数方法,可以自动寻找那些对分类有较好区分能力的支持向量,由此构造出的分类器可以最大化类与类的间隔,具有较好的推广性能和较高的分类准确率,研究了将支持向量机理论用于纹理分类识别的方法,实验结果表明,该方法比传统的基于BP神经网络的识别方法识别准确率高。
Support Vector Machine（SVM）is a model -classification machine based on theories of static learning. It can automatically find out support vectors that have better calssification ability through the risk minimization principle and kernel function. So the classification machine can maximize the interval of each genus and have higher accuracy. Texture recognition can be regarded as an impending problem among different textures and their characteristics. This paper applies SVM to texture recognition and classification. Compared to the BP nerve network,much more ideal results are acquired through experiment.
出处
《电脑开发与应用》
2005年第11期11-12,14,共3页
Computer Development & Applications
关键词
支持向量机
纹理识别
BP神经网络
小波变换
support vector machine, texture recognition, BP nerve network, wavelet transform
