In order to improve the accuracy of the application of remote sensing image in plant classification beyond the weakness of only using spectral information

the texture features of

Apocynum venetum

a typical wild plant in Xinjiang

are extracted based on fractal theory from Worldview2 satellite images by means of double blanket coverage model which are further analyzed by Gray Level Co-occurrence Matrix (GLCM) methods for its classification. Maximum likelihood method is used to classify the textures

and the classification results from different moving window sizes (3×3

5×5

7×7 and 9×9) are compared. The results show that overall classification precision increased by 1.21% to 8.63% for the texture classification compared to traditional spectral information classification

the precision based on fractal theory is more than twice as much as GLCM based texture classification. The classification accuracy of

Apocynum venetum

increased to 99.96% when this parameter was combined with fractal-based textures. By contrast

the accuracy reduced by 0.09% to 0.12% with GLCM-based textures

and there is best classification precision when fractal-based texture was extracted with a 5×5 sliding block. Therefore

the textures based on fractal theory could effectively improve the accuracy of Worldview-2 images in plant classification by GLCM and double blanket coverage model.