Remote sensing classification based on hybrid multi-classifier combination algorithm

To improve the precision of remote sensing image classification, hybrid multi-classifier combination method is proposed. Taking the characteristic of abstract level and measurement level into consideration, the optimal sub-classifier, bagging algorithm and the most large confidence algorithm are combined. By using this model, respective advantages of different sub-classifiers are gathered. This method used in Beijing-1 and ETM image classification shows a better enhancement, and also results indicate that the hybrid multi-classifier combination algorithm is an effective algorithm for medium-high precision remote sensing image classification.