In this paper a Hyperspectral Expert Classifier (HEC) based on data-fusion technique was presented. The spectral-spatial contextual image analysis approaches were applied on hyperspectral images, ETM+ images, and GIS data. First, the samples were selected according to the available information to build the reference spectral and calculate the maximum angle after data fusion. The created maps using Spectral Angle Mapping (SAM), GIS data, hyperspectral image, and ETM+ images were used as an input data in HEC. The result showed that the Land-use in the study area could be identified from Hyperion data efficiently. The hyperspectral expert classifier approach is found to have a merit of high classification precision, low computational cost, and without much interference from the users compared with other classifiers. This methodology could easily extended to a large number of classes and used in practical applications (for example mine exploration).
The regularization parameter and the kernel parameters greatly affect the performance of support vector machines (SVM)
models. This paper proposes an evolutionary algorithm (EA) to automatically determine the optimal parameters of SVM
with the better classification accuracy and generalization ability simultaneously. The proposed ESVM model, called
evolutionary SVM or ESVM, was applied to a Land-cover classification experiment in a 840×840 pixels Landsat-7
Enhanced Thematic Mapper plus (ETM+) high-resolution image of Wuhan in Hubei province of China compared with
the conventional SVM model. Experimental results show that the use of EA for finding the optimal parameters results
mainly in improvements in overall accuracy and generalization ability in comparison with conventional SVM. It is
observed that classification accuracy of up to 91% is achievable for Landsat data produced by ESVM.
A simple approach for incorporating tolerant rough sets (TRS) into a multi-class support vector machine (SVM)
classifier for land-cover classification was presented. TRS was used to perform the sample preprocessing of the original
samples set to reduce the uncertainty of sample set and make the influence of the uncertainty from sample set on the final
classification accuracy least. SVM was employed after the TRS preprocessing. An application of the integrated
classifiers using an ETM+ remote sensing image has been presented. The classification results were compared with those
of only-SVM classifier. According to the overall accuracy and the k coefficient, the result of integrated classifier with
TRS and SVM is better than that of only-SVM classifier in the experiment.