Multi-Classifier Adaptive-Weight Fusion Classification of Applying Cloud Model

• 1、School of Remote Sensing and Information Engineering, Wuhan University,Wuhan 430079

Abstract： Classification research based on multi-classifier fusion is a hot topic in the field of remote sensing image classification, and it can provide more accurate classification result than a single classifier. The difficulties are how to select the target classifiers and how to assign weights to fuse the classifiers. This paper proposes a multi-classifier adaptive-weight fusion classification method based on cloud model. The determination of adaptive weights is divided into two stages. In the first stage, according to the overall distribution of all the samples and the individual characteristics of a single sample in each feature space, the weight of the samplRemote Sensing Image Multi-Classifier Adaptive-Weight Fusion Classification of Applying Cloud Modele is calculated based on cloud model. In the second stage, according to the distribution of output vector of each classifier, the weight of each output node of the classifier is calculated. Every sample in each output node of each classifier is allocated an adaptive weight. Some classification experiments are made by using Landsat TM image to evaluate the proposed method, and classification accuracies are compared quantitatively among different single classifiers and fusion classifiers. The experiment results show the proposed method is effective and can improve the classification accuracy of remote sensing image.

Keywords： multi-classifier fusion adaptive weight cloud model RBF neural network classifier remote sensing image classification