摘要
传统的中国画和书法是我国的艺术瑰宝。随着数字技术的迅速发展,越来越多的传统中国书画作品被数字化,如何快速有效地检索这些数字图像已成为一个热门的研究课题。如果能够准确地将中国画和书法图像首先进行二分类,将为中国书画图像的进一步检索和分类打下坚实的基础。提出了一种基于主体颜色特征的中国传统书画图像的二分类方法。该算法首先对书画图像中的留白区域进行检测,然后将其去掉,因为历史久远,这些留白区域含有过多的噪声;其次,从处理后的书画图像中提取灰度特征作为二分类的基础;最后,利用这些特征训练分类器,并使用训练好的分类器对中国画和书法图像进行二分类。实验结果表明,该算法达到了比较理想的中国书画图像二分类结果。
Traditional chinese painting（TCP） and calligraphy is unique forms of art. With the rapid development of digital technology,more and more TCP and Calligraphy works are digitized. How to effectively retrieve these images becomes a hot topic. If we first classify the TCP and Calligraphy images, this will be a solid foundation for retrievaling those images. We proposed an improved classification method of those images. ＇ Liubai area was detected firstly, and removed it from the images, because these regions contain noise information which will make the classifation results inaccurate. The second step was to extract feature from those images. At last, the features were used to training the Support Vector Machine（SVM） model. And the trained model was used to classifying the TCP and Calligraphy images. The classification result shows this method has better effect.
出处
《计算机科学》
CSCD
北大核心
2012年第3期256-259,共4页
Computer Science
基金
国家自然科学基金(60972145
61070120)
北京市属高等学校人才强教计划资助项目(PHR200907120)资助
关键词
中国画图像
中国书法图像
支持向量机
分类
Chinese painting images, Chinese calligraphy images, SVM, Classifation
