Semi-supervised land cover classification of remotely sensed data using two different types of classifiers

We propose a semi-supervised method of land cover classification for remotely sensed multispectral data. The method is useful especially when the number of training data is small and restricted. The method derives the additional training data out of the object image by using the results of two different types of classifiers. We extract the pixels in which the results of two classifiers were coincide with each other and use them as the additional training data in the classification. By the results of experiments, in which we used maximum likelihood method and AdaBoost for the two classifiers, we confirmed that the algorithm is effective to improved the accuracy of classification.