Implementation of hybrid RS-ANN for spatial image classification

摘要

The main objective of the spatial image classification is to extract information classes from a multiband raster spatial image. The network structure and number of inputs are the key factors in deciding the performance and accuracy of the traditional pixel based image classification techniques like Support Vector Machines (SVM), Artificial Neural Networks (ANN), Fuzzy logic, Decision Trees (DT) and Genetic Algorithms (GA). In this paper, a new hybrid approach is proposed and implemented to improve neural network classification performance that uses rough sets approach for feature selection of image pixels. Code is developed for the implementation of the proposed Rough Set based Artificial Intelligence Neural Network (RS-ANN) technique using JAVA SE JDK, JRE 8, Wolfram Mathematica, R Language Environment version 3.3.1 and R Studio IDE version 1.0.44. The implementation of the tests is done with 20 image instances. It is evident from the results that more number of image instances has less percentage of error than the average error of the corresponding test. The maximum accuracy of the proposed algorithm is 90% which indicates a high accuracy of the proposed hybrid RS-ANN model for spatial image classification.