【摘要】 针对高光谱图像分类中分类精度有待提升和时间复杂度高的问题,论文提出基于核超限学习机的主动学习并用于高光谱遥感图像分类,与几个常见的算法Mclu_Elm,Mclu_Nb,Mclu_SVM和Mclu_Knn进行了对比验证,结果表明,论文提出的方法不仅具有较强的泛化能力,还能极大地缩短响应时间,因此,论文提出的方法适用于实时应用领域。更多还原

【Abstract】 Aiming at the problem that the classification accuracy in hyper-spectral image classification needs to be improved and the time complexity is high,this paper propose an active learning based on the kernel extreme learning machine(mclu-KELM)and is used for the classification of hyper-spectral remote sensing images,which is compared with several common algorithms mclu_ELM,mclu_NB,mclu_SVM and mclu_KNN. The proposed algorithm in this paper not only has strong generalization ability,but also can greatly shorten the time,therefore,the algoritm of this paper is suitable for the field of real-time application.更多还原