Abstract: In the application of remote sensing image classification,the classification accuracy and efficiency of the classification results obtained by using different kinds of classifiers to operate on the original image are different.In this paper the multi classifier fusion classification experiment are designed and completed.The experiment can combine the advantages of different single classifiers with appropriate methods,so as to obtain better classification accuracy and efficiency than single classifier Method.On this basis,the voting principle is used to design the multi classifier of abstract level fusion and complete the experiment.The design results show that the classification accuracy of multi classifier fusion is higher than that of single classifier.