CAJ | 학술논문

图像分类方法的研究对图像的检索、识别等具有重要意义。局部支持向量机能够有效利用样本的局部信息，与标准支持向量机等分类算法相比具有较高的分类精度。邻域核函数是一种能够有效反映邻域信息变化状况的核函数。将邻域核函数与局部支持向量机相结合，提出一种基于邻域核函数的局部支持向量机，并将该算法应用于图像分类识别中。为验证算法的有效性，使用树木图像数据集进行测试实验。实验结果表明，将邻域核函数应用于局部支持向量机，能够有效提高图像的分类精度。
도상분류방법적연구대도상적검색、식별등구유중요의의。국부지지향량궤능구유효이용양본적국부신식，여표준지지향량궤등분류산법상비구유교고적분류정도。린역핵함수시일충능구유효반영린역신식변화상황적핵함수。장린역핵함수여국부지지향량궤상결합，제출일충기우린역핵함수적국부지지향량궤，병장해산법응용우도상분류식별중。위험증산법적유효성，사용수목도상수거집진행측시실험。실험결과표명，장린역핵함수응용우국부지지향량궤，능구유효제고도상적분류정도。
Local Support Vector Machine can effectively use the local information of samples. Compared with the standard Support Vector Machine, it has higher classification accuracy. The neighborhood kernel function can make effective use of neighborhood change information. This paper proposes a novel Local Support Vector Machine algorithm based on neighborhood kernel function and used it in image classification. In order to verify the validity of the method, this paper did tests using image data. The experimental results show that, this novel LSVM has higher classification accuracy than SVM and LSVM. So it can effectively improve the accuracy of image classification.