［Purposes ］ Image vegetation classification based on traditional vegetation index is a relatively mature method，but there are often problems of missing and misclassification in practical applications.This paper aims to solve the above problems with a new image vegetation classification method.［Methods］ This paper analyzed the shortcomings of the image vegetation classification method based on traditional vegetation index from the geometric significance of vegetation index in red -green -blue （RGB） color space and revealed the color information distribution law of vegetation in RGB space.It proposed an image vegetation classification method based on the RGB three -axis ellipsoid model and determined the parameters of its model through principal component analysis （PCA） and statistical calculation.［Findings］ Under the same sampling conditions，compared with the image vegetation classification method based on vegetation index，the image vegetation classification method based on the RGB ellipsoid model yields better results.［Conclusions ］ The method in this paper improves the problem of missing and misclassification in image vegetation classification and gives a new method for image vegetation classification.