Interspecies classification of mangroves based on data dimension reduction using hyperspectral UAV images
Shenzhen Futian Nature Reserve was studied,and a hyperspectral unmanned aerial vehicle(UAV)remote sensing system was applied to conduct research on remote sensing of mangrove resources.Remote sensing information extraction and interspecies fine classification models of mangroves were established.Firstly,a decision tree classification method based on expert knowledge was adopted,and normalized differential vegetation index(NDVI)was used to distinguish mangroves from other features.Then,minimum noise fraction(MNF)transformation was performed on the mangrove data,and the optimum index factor(OIF)formula was used to analyze the first 11 wave bands after the MNF transformation,so as to get the optimal wave band combination.Finally,the pixel-based minimum distance method and support vector machine method were used to perform interspecies classification of mangroves based on the optimal wave band combination.The results show that the decision tree classification method based on expert knowledge can effectively extract the area with mangroves.The pixel-based support vector machine method performs better in interspecies classification,and the overall classification accuracy can reach over 99%.The mangrove extraction and tree species classification models based on expert knowledge are applicable to the interspecies classification and mapping of mangroves.
国家科技期刊平台
NSTL
万方数据
