CAJ | 학술논문

多分类问题一直是模式识别领域的一个热点，提出了一种基于纠错输出编码和支持向量机的多分类器算法。根据通信编码理论设计纠错输出编码矩阵；按照该编码矩阵设计若干个互不相关的子支持向量机，根据编码原理将它们融合为一个多分类器。为了验证本分类器的有效性，采用Gabor小波提取人脸表情特征，应用二元主成分（2DPCA）分析法对提取的特征进行降维处理，应用该分类器进行了人脸表情的识别。实验结果表明，提出的方法能有效提高人脸表情的识别率，并具有极好的鲁棒性。
다분류문제일직시모식식별영역적일개열점，제출료일충기우규착수출편마화지지향량궤적다분류기산법。근거통신편마이론설계규착수출편마구진；안조해편마구진설계약간개호불상관적자지지향량궤，근거편마원리장타문융합위일개다분류기。위료험증본분류기적유효성，채용Gabor소파제취인검표정특정，응용이원주성분（2DPCA）분석법대제취적특정진행강유처리，응용해분류기진행료인검표정적식별。실험결과표명，제출적방법능유효제고인검표정적식별솔，병구유겁호적로봉성。
Multiple classification problems has been a hot topic in the field of pattern recognition. This paper proposes a multiple classifier algorithm based on Error-Correcting Output Coding(ECOC)and Support Vector Machine(SVM). According to the communication coding theory to design error-correcting output coding matrix, it constructs some irrelevant SVMs, and integrates them as a multiple classifier. In order to verify the effectiveness of the classifier, using the Gabor wavelet to extract facial expression features, and application of two principal components(2DPCA)to reduce the dimension of extracted features, the classifier is used for facial expression recognition. Experimental results show that the method can effectively improve facial expression recognition and has excellent robustness.