In this paper

we present the results of our research to evaluate the accuracy of the back propagation neural network method to classify forest vegetation using a 27 July 2001 Landsat 7 ETM+ image of the Manhanshan Forestry Center.The type and quantitative accuracy of the back propagation neural network are compared with the maximum likelihood

the simple and the complex unsupervised classification methods.The total cover type accuracy of back propagation neural network classification is 70.5%

the total quantity accuracy is 84.65%

and the KAPPA coefficient is 0.6455.Our results indicate that the total type accuracy increases 10.5%、32% and 33% respectively compared to the other three classification methods.Total quantitative accuracy increases 5.3%.It is evident that the classification quality of the back propagation neural network is better than the other methods.Therefore

the back propagation neural network is an effective and accurate method of classifying forest vegetation.