In this paper, a multiple classifiers system is utilized to improve the classification performance by integrating the outputs of several classifiers. A weighting policy is applied to fuse knowledge acquired by classifiers to arrive at an overall decision that is supposedly superior to that attainable by any one of them acting alone. The most popular weighting policies, Bagging and Boosting, are introduced. By adopting the concept of confidence index, which accounts for the ambiguities among classes, the modified Bagging and Boosting weighted multiple classifiers systems are proposed. The classification performances of utilizing the original and modified Bagging and Boosting weighted multiple classifiers systems to the application of remote sensing image classification are demonstrated and compared. Experimental results show that the classification accuracy is considerably improved by making use of the multiple classifiers system. In addition, the multiple classifiers systems of using modified Bagging and Boosting algorithms are superior to those of using original Bagging and Boosting algorithms. Moreover, the effectiveness of using confidence index is obviously, especially when the ambiguities among classes are high.
展开▼
