Multi-channel/multi-sensor image classification using hierarchical clustering and fuzzy classification

Various satellite-based sensors currently provide different information about the Earth´s surface. Recently, there has been increasing interest in the use of multi-sensor data for remote sensing applications. The purpose of this study is to classify the land-cover using remotely-sensed data from multiple sources. Most of statistical classifier requires the knowledge of the number of classes and the class parameters, which are not previously known in practice. The process of collecting the information necessary for the classification, however, is very expensive. This study has developed to unsupervisedly estimate the number of classes and the parameters of defining the classes in order to train the classifier. A hierarchical clustering procedure and local fuzzy classification have been employed to find the sample classes that well represent the ground truth. The maximum likelihood classifier has then used the sample classes. The combination of image bands associated with multiple channels/sensors has been selected based on the signal to noise ratio, which is the ratio of difference in class-signal and signal-dependent noise