论文部分内容阅读
针对互花米草的爆发式增长对沿海滩涂生物多样性和生态稳定带来了巨大的生态威胁的问题,该文以闽江河口互花米草和其他3种湿地植物的室内叶片高光谱数据为例,探讨互花米草与其伴生植物是否具有光谱可分性。采用一种分层分析方法对实测高光谱数据降维并选择出识别互花米草的最佳波段。首先,利用ANOVA对光谱数据降维,选择出互花米草与其他湿地植物光谱具有显著性差异的波段;其次,使用CART算法对ANOVA降维后具有显著差异的高光谱数据进一步降维,找到识别互花米草潜在的最佳波段;最后,利用J-M距离评估CART选择波段的可分性。结果表明:互花米草与其他3种湿地植物具有光谱可分性,其J-M距离均高于1.9;基于CART算法的入侵种互花米草的识别精度平均达到96.7%,高于传统方法的识别精度。该文成果将为航空或航天高光谱遥感监测互花米草入侵区提供参考。
In response to the explosive growth of Spartina alterniflora on the coastal biological diversity and ecological stability has brought a huge ecological threat to the issue of the Minjiang estuary Spartina and three other wetland plants indoor leaf hyperspectral data As an example, to explore whether Spartina alterniflora and its accompanying plants have spectral separability. A hierarchical analysis method is used to reduce the dimension of the measured hyperspectral data and select the best band to identify Spartina alterniflora. Firstly, the dimension of the spectral data is reduced by ANOVA, and the band with significant difference between the Spartina alterniflora and other wetland plants is selected. Secondly, using the CART algorithm to further reduce the dimension of the hyperspectral data after dimensionality reduction by ANOVA, Identify the potential best band for Spartina alterniflora; finally, evaluate the separability of the CART selected band using JM distance. The results showed that the Spartina alterniflora and other three wetland plants had spectral separability, and the JM distances were all higher than 1.9. The identification accuracy of the invasive species Spartina alterniflora based on the CART algorithm was 96.7%, which was higher than that of the traditional methods Recognition accuracy. The results of this paper will provide references for the monitoring of Spartina alterniflora by airborne or space hyperspectral remote sensing.