关键词: 图像分类, 特征融合, 空间金字塔, 稀疏编码, 支持向量机

Abstract: Abstract: Concern the problem that the traditional sparse coding image classification algorithms extracted the single type feature, ignored the spatial structure information of the image and lacked the topological structure information of the feature. In order to solve the problems, the multi-scale feature fusion based Hessian sparse coding algorithm was proposed. Firstly, the image was divided into sub-regions with multi-scale spatial pyramid; Then, the Histogram of Oriented Gradient and Scale-invariant feature transform were effectively merged in each subspace layer; In order to make full use of the characteristic topology information, the second order Hessian energy function was introduced to the traditional sparse coding target function as a regular term; Finally, the support vector machine was used to classify. In the Scene15 experiment, the accuracy of HSC was 3%-5% higher than that of Locality-constrained Linear Coding, and it was 1%-3% higher than that of support discrimination dictionary learning. In the tine-consuming experiment on Caltech101, HSC is about 40% less than the Multiple Kernel Learning Sparse Coding. The experimental results show that the HSC algorithm can effectively improve the classification accuracy, and the efficiency is better than the contrast algorithm.

Key words: Keywords: image classification, feature fusion, spatial pyramid matching, sparse coding, support vector machine