论文部分内容阅读
机器学习方法在高光谱遥感影像分类中广泛应用,本文使用新型的极限学习机(Extreme Learning Machine,ELM)进行高光谱遥感影像分类,针对ELM中正则化参数C和核参数σ,提出以萤火虫算法(Firefly Algorithm,FA)进行优化。首先,采用萤火虫算法进行高光谱遥感影像的波段选择,以便降低维数;然后,利用萤火虫算法以分类精度最大化为准则对ELM的参数组合(C,σ)进行寻优;最后,利用参数优化后的ELM分类器,对3个不同传感器的高光谱遥感影像进行分类。实验中将新型的萤火虫算法与遗传算法(Genetic Algorithm,GA)和粒子群算法(Particle Swarm Optimization,PSO)进行了对比,并将ELM的性能与支持向量机(Support Vector Machine,SVM)方法作对比。结果表明,FA优化方法优于传统的GA和PSO优化方法,ELM方法的效果在训练时间和分类准确率2个方面都优于SVM方法。实验说明,本文提出的方法具有较好的适用性和较优的分类效果。
The machine learning method is widely used in the classification of hyperspectral remote sensing images. In this paper, hyperspectral remote sensing image classification is carried out by using a new extreme learning machine (ELM). For the regularization parameter C and nuclear parameter σ in the ELM, a firefly algorithm (Firefly Algorithm, FA) to optimize. Firstly, the firefly algorithm was used to select the band of hyperspectral remote sensing image in order to reduce the dimensionality. Secondly, the firefly algorithm was used to optimize the parameter combination (C, σ) of ELM with the maximization of classification accuracy. Finally, The latter ELM classifier classifies the hyperspectral remote sensing images of three different sensors. In the experiment, the new firefly algorithm was compared with Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), and the performance of ELM was compared with Support Vector Machine (SVM) . The results show that the FA optimization method is superior to the traditional GA and PSO optimization methods, and the ELM method is superior to the SVM method in both training time and classification accuracy. Experimental results show that the proposed method has good applicability and better classification results.