In this article

OMIS hyperspctral data was used to study feature selection ability of DT（Decision Tree） algorithm and the impacts of feature selection on DT.The DT was compared to three designed feature selection methods（SEP

MDLM and RELIEF） based on feature selection results and classification accuracy in which three different methods（ML、 BPNN and 1-NN） were applied.Moreover

the impacts of the three designed feature selection methods on DT classification results at different training sample sizes were analyzed.Results indicated that DT was a good feature selection method.After feature selection

DT algorithm outputted to those classification trees that used fewer features（average decrease was 43.36%）

had fewer tree nodes（average increase was 18.61%）

and had higher classification accuracy（average increase was 0.35%）.When the training sample size was small

accuracy improvement was the most significant and meanwhile the tree size scarcely changed.