A new approach to evaluate the accuracy of image classification result $R'

In order to either ascertain the best model for remote sensing image classification or evaluate the accuracy of the classification result, the paper introduced R' into the image classification precision evaluation. Aiming at the remote sensing image classification characters, the R' was improved and generalized. On the basis of R' theory analysis, the method was tested by classifying the actual image and assessing the classification result. The experiment showed that R' is a new and efficient method for images' classification model efficiency assessment and it will be generally used in the future