Spectral-Spatial Hyperspectral Image Classification Using Superpixel and Extreme Learning Machines

摘要: We propose an efficient framework for hyperspectral image (HSI) classification based on superpixel and extreme learning machines (ELMs). One can be regarded as a small region consisting of number pixels with similar spectral characteristics. The novel utilizes to exploit spatial information which improve accuracy. Specifically, we first adopt segmentation algorithm divide the HSI into many superpixels. Then, features superpixels are extracted by computing mean within each superpixel. feature combine Finally, ELMs is used determine class label Experiments two real HSIs demonstrate outstanding performance proposed method in terms accuracies high computational efficiency.