A remote sensing image classification method based on sparse representation

With the development of remote sensing image applications, sparse-based representation classification approaches have been investigated for better classification accuracy. This paper introduces an improved classification method based on sparse representation by representing the test samples through a dictionary. The key components of our proposed method rely on the feature dictionary construction, sparse representation and image reconstruction. The dictionary is obtained by training samples according to their class for a sparse linear combination. The sparse representation for the image is expressed as sparse coefficients by solving an optimization problem. We describe the method of constructing a dictionary by computing a best matrix to represent all data vectors. We also describe the algorithm used to solve for the sparse representation. Finally, we discuss the way of using the sparse vector to reconstruct the image for classification. In the experiments, the proposed method is applied to two real high spatial resolution images for the classification in comparison to Backpropagation Neural Network, Support Vector Machine, Classification and Regression Trees and K-means. The experimental results show that the proposed method performs better than the benchmark methods in terms of classification accuracy.