Abstract: Chinese Painting（CP） is the gem of Chinese traditional arts，browsing it based on the semantic is necessary.The semantic of CP reflect mainly at color and shape.The paper according to the own features of CP，discusses the algorithm of color and shape，makes the feature of color and aim shape together，builds a new feature vector，analyzes the relationship between high-level semantic and multi low-level features based on the CP image.Semantic classification is performed by SVM，the experiment indicates that this method improves the image semantic classification precision.

Key words: image of Chinese Painting, feature extraction, SVM, semantic classification

摘要： 中国画作为中国传统文化艺术的瑰宝，根据语义对国画图像进行检索是必要的。国画的语义主要反映在颜色和形状。依据国画自身的特点，研究了颜色和形状的特征提取算法，融合图像的颜色和目标的形状特征，构建了一种新的特征向量，分析了国画图像的多维低阶特征与高阶语义之间的相关性，采用支持向量机实现语义分类，实验结果表明该方法提取的特征向量稳定，能得到较高的分类精度。

关键词: 国画图像, 特征提取, 支持向量机, 语义分类