Extracting croplands in western Inner Mongolia by using random forest and temporal feature selection

Reliable information of croplands has useful implications for agriculture. Based on random forest classifier, a cropland extraction method was developed. Multi-temporal image data of Landsat 8 were used for classifying crop and non-crop vegetation, since these data contain useful information. However, there is also large redundancy. To solve this problem, a new feature selection method was proposed in this paper. The primary innovativeness resides in a temporal feature selection criterion. A classification experiment was carried out to validate the proposed technique. The results showed that the proposed method can decrease feature redundancy with little cost on classification performance.