中文摘要：

针对单分类器没有充分考虑数据集的特征而不能很好地完成分类识别，提出了一种基于集成学习技术的SVM集成的图像分类方法。该方法是在基于较为流行的词袋（Bag—of—words，BOW）模型的图像分类方法的基础上，利用训练生成的不同SVM分类器分类测试图像，并将分类结果采用集成学习算法进行集成。分别采用传统的BOW模型的图像分类方法和本文提出的方法进行分类实验，实验结果表明采用SVM集成的图像分类方法明显提高了分类精度，具有一定的稳健性。

英文摘要：

Single classifier can＇ t fully catch the dataset＇s features to effectively implement classification and recognition, so a SVM ensemble classifica- tion method based on ensemble learning technology is proposed. The method is based on the popular image classification method of BOW. It uses the dif- ferent trained SVM classifiers to classify the testing images and uses ensemble learning algorithm to integrate the classification results. The traditional BOW image classification method and the proposed method are used to do images classification experiment separately. The experimental results show that the proposed method can effectively implement the image classification and improve classification precision with a certain degree of robustness.