Abstract: SURF（Speeded Up Robust Feature） has excellent description ability for local features, but isn’t strong for describing global features. This paper proposes an image classification algorithm based on the combination of SURF and global features. The SURF vector sets are extracted, and the vector sets are reduced to a single high dimensions feature vector by employing the random histogram algorithm. The HSV（Hue, Saturation, and Value） color histogram is extracted. These two features are classified with SVM（Support Vector Machine）, respectively. The two classification results are integrated by the algorithm of high-level cue integration to get the ultimate classification result. The experimental results demonstrate that the proposed algorithm greatly improves the accuracy of image classification.

Key words: Speeded Up Robust Feature（SURF）, global feature, random histogram, Support Vector Machine（SVM）, feature fusion

摘要： 针对SURF对图像局部特征具有极好的描述能力，但对于全局特征描述能力不强的缺点，提出将SURF和全局颜色特征相融合的图像分类算法，提取图像的SURF特征向量集，并利用随机直方图算法将该向量集进行数据归约成单一高维特征向量；提取图像HSV颜色直方图;分别利用支持向量机（SVM）对这两种特征进行分类;将两个分类结果进行高层特征融合得到最终分类结果。实验结果表明，该算法显著提高了图像分类的准确率。

关键词: 快速鲁棒特征（SURF）, 全局特征, 随机直方图, 支持向量机, 特征融合