摘要
将支持向量机应用到纹理识别领域,提出了一种基于支持向量机和小波变换的新型纹理识别方法。该方法用小波变换各子带图像共生矩阵参数、分析窗口大小、像素均值和像素标准差等参数作为纹理特征,解决了描述不同尺度纹理的难题。以多类支持向量机作为分类器,用输出纠错码把二分类器扩展到多类,提高了分类器的泛化能力。在包含25类单色自然纹理的图像库上进行识别试验,结果表明,该方法识别错误率小于10%,识别正确率比传统的贝叶斯等方法提高了2%左右,获得了更高的识别正确率,且推广性更好。
A novel texture recognition method based on Support Vector Machines （SVMs） and wavelet transform is proposed. To alleviate the problem of characterizing different scales of textures, the co-occurrence parameters, window size, mean and standard deviation of different level discrete wavelet transform （DWT） images are used as texture features. In particular, we apply SVMs for texture recognition, using Error Correcting Output Codes （ECOC） to extend binary SVMs to multi-classifiers. Compared with conventional Bayes classifier, SVMs have more accurate recognition rates tested on with Ponce texture database including 25 kinds of monochromatic natural texture. Experiments indicate that the error rate of this method is less than 10% and its recognition accuracy is improved by about 2% than that of conventional Bayes classifiers.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2005年第8期48-51,共4页
Opto-Electronic Engineering
基金
国家自然科学基金资助项目(60272014)
关键词
纹理识别
支持向量机
离散小波变换
共生矩阵
模式识别
Texture recognition
Support vector machines
Discrete wavelet transform
Co-occurrence matrix
Pattern recognition
