中文摘要：

提出一种采用卷积神经网络对自然图像和文档扫描图像进行分类的方法，通过卷积和池化操作提取两类图像具有高区分度的特征，融合后得到分类判决结果。实验结果表明，所提出的分类方法在SKL图像库上分类精度超过93％。图像预处理对模型的精度以及模型训练收敛所需时间具有积极效果，经过图像预处理后训练的卷积神经网络模型对图像文字大小和图像格式顽健。

英文摘要：

Using convolution neural network which though convolution and pooling extracting features of high dis- tinguish ability and then make fusion for classification of natural images and scanned documents. Experimental re- sults show that the classification accuracy of the proposed classification method is more than 93% on the SKL image database. The model is highly robust to font sizes and image formats. Through contrast experiment validated that preprocessing of image has a positive effect on the accuracy of the model and the time cost on training.