Multi-sensor data classification with integrated neural network method

Multi-sensor and multi-temporal data classification with combined neural network methods has been implemented in this paper. Pre-classification has been carried out by using Kohonen neural network processing technique, which allows the simultaneous use of data acquired by different sensors with no additional hypothesis on data distribution. Especially some of the experiments reported that combination of different types of sensors provide satisfactory results than using only one sensor data. This paper also shows the multi-sensor data classification is better than single sensor data classification. The proposed method is illustrated with the aid of examples of satellite image of Daejeon, Korea. Features obtained from the pre-classification are used as input to the error backpropagation algorithm for training, and classification. The results of the proposed method with multilayer perceptron (MLP) classifier are compared with the results of the maximum likelihood classifier (MLC).