Some Experiments on Peer Group Pixels Based Spatial Classifier

This paper empirically compares two algorithms for classifying remotely sensed images: The minimum-distance classifier and peer group pixels based spatial classifier. The minimum-distance classifier is a per-pixel classifier. The spatial classifier incorporates contextual information by using the average of the pixels from a peer group of each pixel and classifying this pixel into one of several predefined classes based on the average. Experiments on noisy images were performed in this study. Noisy images will be smoothed and then use the perpixel based classifier for the comparison. The results are assessed by visual inspection.