Object-oriented Fuzzy Classification For Remote Sensing Image

The traditional pixel-based remote sensing image processing approaches exploit the features of abundant spatial information and explicit difference of spectrum between different objects. Applying the traditional classifying algorithms to the high resolution remote sensing image with few spectral bands will cause the low classification rate, redundancy spatial data, and output image with pepper-salt noise.Fuzzy classification is a technique that basically translates feature value of arbitrary range into fuzzy values between 0 and 1, indicating the degree of membership to a specific class. By translating feature values into fuzzy values, fuzzy classification can standardize features and allows the combination of features, even of very different range and dimension. Fuzzy classification also provides a transparent and adaptable feature description.Object-based image classification, which is based on fuzzy logic, allows the integration of a broad spectrum of different object features such as spectral values, shape, and texture. Such classification techniques, incorporating contextual and semantic information, can be performed using not only image object attributes, but also the relationship among different image objects.This research has main two steps. First step is segmentation. During this process, how to choose fine scale is very important. Second step is classification. In this step, two factors are very essential for classification. Creating feature space is one of the factors, another is fuzzy logic. After these two steps, we can extract some classes that we are interested in.