A Novel Robust Classification Method for Ground-Based Clouds

Though the traditional convolutional neural network has a high recognition rate in cloud classification, it poor robustness classification with occlusion. In this paper, we propose novel scheme for which networks are used feature extraction and weighted sparse representation coding is adopted classification. Three such algorithms proposed. Experiments carried out using multimodal ground-based dataset results show that case of occlusion, accuracy proposed methods can be much improved over network-based algorithms.