Combination of hard and soft classification method based on adaptive threshold

This paper has presented a hard and soft classification model that based on hard and soft classification technique to mapping vegetation distributions. It chose SVMs class image as hard classification model and LSMM results as soft classification model. Through a new adaptive threshold algorithm which could define pure and mixed regions of vegetation automatically to combine hard classification results and soft classification results. In the agricultural landscapes of Southeast Beijing City, results from the proposed model were assessed at a range of spatial scales. Results of vegetation distributions were compared with hard classification model and soft classification model with RMSE. Accuracy assessment showed that hard and soft classification model could get better results.