摘要
引入了将模糊理论融入SVM的改进理念,即FSVM,并实现了一种利用FSVM作为分类器且对图像逐层进行分类直至情感语义层面的分类系统,其难点在于建立从图像的低阶图像特征到高阶语义特征之间的映射关系,以及如何选取适合的隶属度函数来确立测试图片的具体语义类别。实验结果表明,本系统在图像情感识别中确实具有简单、快速、高效等特点,从而证明本系统将图像的语义分类提升到情感层面是成功的。
This paper introduced FSVM,whieh introjects fuzzy theory to SVM, achieves a classification system which classifies image layer by layer to affective semantic level by FSVM,and proposed one kind of image affective semantics classification method. The difficulty is to establish a mapping from image features to image affective semantics and how to select fitting membership function to test image semantic class. The experimental result shows that the system is simple, fast, effective, and so on, therefore our system is proved to be successful in promoting the image semantic classification to affective semantic level.
出处
《计算机科学》
CSCD
北大核心
2009年第8期288-290,299,共4页
Computer Science
基金
国家自然科学基金(60773004)
山西省自然科学基金(2006011030
2007011050)资助
