Abstract: Deep learning shows excellent performance in image recognition and natural language processing. This paper presents a learning method which is based on Convolution Neural Network（CNN） to classify remote sensing images. Aiming at the problem of low efficiency caused by too many sampling windows in remote sensing image classification, this paper proposes a novel method based on region to improve the classification efficiency. There are two steps in the classification process. The first step of the proposed algorithm is to classify the input images by convolution neural networks. Then, SVM is used to classify the results of convolution neural networks. The proposed algorithm is evaluated by being compared with the results of other existing methods. Experimental results show that proposed algorithm can achieve higher classification accuracy.

Key words: Convolution Neural Networks（CNN）, image segmentation, image classification, transfer learning

摘要： 深度学习的方法在图像识别和自然语言处理等方面展示了优异的性能。将卷积神经网络（Convolution Neural Network，CNN）用于高分辨率遥感影像分类。针对CNN用于遥感影像分类使用固定大小窗口遍历时，影像采样窗口数量过多，导致的分类效率低下问题，提出一种基于影像区域特性的采样窗口确定方法，提高分类效率。影像分类包括两个阶段：首先，利用卷积神经网络得到的特征对影像进行分类；然后，采用支撑向量机对第一步分类由于特征区分性不足造成的错分地物类别进行再分类。采用具有不同特性的遥感影像对所提方法进行了验证，实验结果表明，同现有的特征表示和分类方法相比，该方法的性能有明显改善。

关键词: 卷积神经网络（CNN）, 影像分割, 影像分类, 迁移学习