基于稀疏自动编码器和支持向量机的图像分类

Image Classification Based on Sparse Autoencoder and Support Vector Machine

针对单层稀疏自动编码器在特征学习时容易丢失深层抽象特征,特征缺乏鲁棒性的缺点,提出一种新的基于稀疏自动编码器和支持向量机的图像分类方法。构建深度稀疏自动编码器对图像逐层学习并自动提取每层特征,根据特征集权值重组法得到每层特征权值和重组特征集。将遗传算法强大的全局搜索能力和支持向量机分类优势结合,高效、准确的完成图像分类。实验结果表明,该算法能自动地学习图像深层特征,重组特征集具有较高的特征识别力,有效地提高了图像分类准确率。

A new algorithm of image classification based on the sparse autoencoder and the support vector machine was proposed in view of the drawbacks that the single layer sparse autoencoder for feature learning is easy to lose the deep abstract feature and the features lack the robustness. The deep sparse autoencoder is constructed to learn each image layer and the feature of each layer is automatically extracted. The each feature weights and the reorganized set of feature are obtained according to the feature weighting method. By combining the strong global search ability of genetic algorithm and the excellent performance of support vector machine, the image classification is completed efficiently and accurately. The experimental results show that the proposed algorithm can automatically learn the deep feature of the image, and the reorganized feature has high feature discrimination ability, which effectively improves the accuracy of image classification.