论文部分内容阅读
本文为验证SVM对高维特征的适应性和可靠性,针对不同特征提取方法与特征组合,以国产OMISⅡ传感器获得的北京昌平地区高光谱遥感据为例,对SVM分类器中特征维数对分类准确率的影响进行了试验,通过对主成分分析、最小噪声分离算法、相关系数分组后特征提取、导数光谱等的分析,表明SVM分类器的分类精度随着特征维数波动,其中主成分分析降维后提取的特征具有用于分类能够获得最高的准确率。通过与最大似然法和光谱角制图分类算法的比较,说明在同样的特征输入情况下SVM分类算法分类的准确率高于最大似然法和光谱角制图分类器。
In order to verify the adaptability and reliability of SVM to high-dimensional features, aiming at the different feature extraction methods and feature combinations, the paper takes the Beijing Changping area hyperspectral remote sensing data obtained by domestic OMISⅡ sensor as an example, The main results are as follows: 1. The main components analysis, the least noise separation algorithm, the feature extraction after the correlation coefficient grouping and the derivative spectra were analyzed. The results show that the classification accuracy of SVM classifier fluctuates with the feature dimension, Features extracted after dimension reduction have the highest accuracy for classification. By comparing with maximum likelihood method and spectral angle classification algorithm, it shows that SVM classification algorithm has higher classification accuracy than maximum likelihood method and spectral angle classifier with the same feature input.