Abstract:

In order to improve the classification accuracy of hyperspectral image data and reduce its dependence on a large number of data sets, we propose an improved feature extraction scheme based on weighted fuzzy C-means algorithm. The weighted fuzzy C-means algorithm is applied to assign different weights to each feature, thus ensuring the extracted features contain more effective information so as to reduce the number of training data sets without reducing the amount of information needed for classification. Experimental results show that compared with the prototype spatial feature extraction method, the proposed method is stable and has a higher classification accuracy under small data sets.