HYPERSPECTRAL SENSOR DATA FUSION AT DECISION LEVEL USING SUPPORT VECTOR MACHINE

Y.Sai Praveen, A.Kiranmai, K.Nikitha, V.Sowmya Devi

Abstract: SVM is the major feature extraction technique extensively used in the hyperspectral sensor. First, analysis aimed at understanding and assessing the potentialities of SVM classifiers in data fusion at decision level is presented. This is very significant particularly for improving the algorithms for classification of remote sensing data from different types of sensors operating in hyperspectral or panchromatic mode. The major problem occurs with high dimensionality of hyperspectral sensor data and nonlinear characterization, SVM can overcome these problems and deliver a better result when compared with other classifiers at decision level. The performance and efficiency of SVM depend upon the problem and different kernel functions used in SVM classifier at decision level data fusion. In this paper, problems in the classification of hyperspectral sensor data are analysed and applicability of decision level data fusion as a part of the enhancement of sensor data capability is studied.

Keywords: Classification of hyperspectral data, artificial neural networks, Support vector machine, Decision-level Fusion.

DOI: https://doi.org/10.15623/ijret.2016.0524005