摘要
针对多项式核或RBF核SVM不能很好地处理图像标注中的数据不平衡问题,提出了一种基于多核函数SVM的图像标注方法,该方法采用多核函数训练过的SVM将基于区域的图像标注问题转化为对非平衡数据分类的问题,进而对图像标签进行分类以获取更符合图像真实含义的标注.实验结果表明,多核函数SVM图像标注性能优于单独使用局部或全局核函数.
To solve the inability of the polynomial kernel or RBF kernel support vector machine（SVM） to handle image annotation data imbalance,an image annotation method based on multiple-kernel SVM is proposed.The method uses SVM trained by multiple kernel to turn the problem of region-based image annotation into the problem of unbalanced data classification.The image tags are then classified to obtain annotation in line with the true meaning of the image.Experimental results show that the performance of multiple kernel SVM image annotation works better than the method based on the part kernel or global kernel function alone.
出处
《昆明理工大学学报（自然科学版）》
CAS
北大核心
2012年第5期43-46,53,共5页
Journal of Kunming University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(70971067)
江苏省科技型企业技术创新资金项目(BC2012201)
关键词
支持向量机
多核函数
图像标签
support vector machine
multiple kernel function
image annotation
