Study on Remote Sensing Extracting Camellia oleifera Forest Based on GF-6 Multispectral Imagery
[Objective]The remote sensing extraction method for Camellia oleifera forests based on GF-6 multispectral imagery was explored to provide references for high-precision recognition and monitoring of C.oleifera forests.[Method]Based on the classification extraction technology of high-resolution optical remote sensing imaging,GF-6 WFV multispectral images and vegetation indices were used to construct a classification model on the basis of random forest combined with spectral features and red edge index features,and then the spectral features,category separability and vegetation index features of C.oleifera forest and other land cover in four phases were compared and analyzed to evaluate the extraction accuracy of the model.[Result]After adding the optimal vegetation index to the 8-band classification based on random forest,the overall classification accuracy reached 95.63%,an increase of 2.96 percentage points compared to the original classification accuracy.The mapping accuracy and user accuracy of C.oleifera forest reached 93.46%and 96.51%,respectively,with an increase of 2.60 percentage points and 3.67 percentage points,respectively.The red edge indices NDVI750 and IRECI could improve the model's ability to distinguish C.oleifera forests from other land types,thereby enhancing the accuracy of extracting C.oleifera forests.[Conclusion]GF-6's wide 8-band multispectral images have the ability to effectively identify C.oleifera forests macroscopically,among which the red-edge band,yellow band,blue band and purple band play an important role in the accurate recognistion of C.oleifera.The use of a random forest 8-band feature classification model adding red edge index NDVI750 and IRECI,and vegetation index NDVI and RVI can effectively improve the accuracy of identifying C.oleifera forest patches.
万方数据
维普
