Multi Spectral Satellite Image Ensembles Classification Combining k-means, LVQ and SVM Classification Techniques

There is no doubt that every classification technique involves advantages and drawbacks like any people thoughts, and as gathering different thoughts aids people to ensure and support for a beneficial decision, the combining of different classification techniques helps to construct strong classification system supplying better results. Ensemble classification techniques are applied today in most image classification fields, and remote sensing image classification is one of those areas giving a growing number of trust with this kind of classification system. In this paper, we used and combined three popular classification techniques using majority vote method. First will be the unsupervised segmentation technique k-means that we follow by SVM (support vector machine) regularization. The last technique will be the supervised neuronal technique LVQ (linear vector quantization). The main contribution of this work is to combine three different kinds of classification methods; the unsupervised k-means technique to get an initial view of the image site that will helps us to choose the best training area for SVM classification system which is the second kind. The last kind is the neuronal supervised LVQ technique. Finally, we used in this specific work a LANDSAT image of a well known south Algerian region which is the Ouargla oasis acquired on December, 20 in 2000.