Rough-set and artificial neural networks based image classification

Spatial image classification meant to the mechanism of extracting meaningful knowledge information classes from spatial images dataset. Many traditional pixel based image classification techniques such as Support Vector Machines (SVM), ANN, Fuzzy methods, Decision Trees (DT) etc. exist. The performance and accuracy of these image classification methods depends upon the network structure and number of inputs. Here, in this paper, we have proposed an step-wise mechanism to significantly improve the classification performance of neural network, that uses rough sets approach for purpose of features/attributes selection of image pixels. The complexity analysis of the proposed algorithm and the comparison of mechanism, presented here, with existing classification techniques based on features over the interest area is carried out.