{Abstract] This paper proposes a modified feature extraction and classification method to classify the electronic house property archives and extract the key information. It makes paragraph features combining up with the local relative pixel distributions. With the template in training, it makes Support Vector Machine(SVM) classifier to automatic storage house property archives image. Experimental results show that this method has better classification effect and stability.%针对房产电子档案分类与关键信息提取的问题,提出一种改进的特征提取与分类方法.将文档图像的段落特征和局部像素分布值特征作为房产文档图像的综合特征,利用模板训练支持向量机分类器,实现对入库的房产档案图像的自动分类.实验结果表明,该方法的分类效果和稳定性较好.
展开▼