Satellite Image Classification Using Image Encoding and Artificial Neural Network

The conceptual variety in the satellite images cause some differences in color ingredients and fine details that usually need to be distinguished for purpose of classification. Image classification is carried out generally using a five to seven color bands that are reflect signitures of the objects found in the landcover. The geographic information system (GIS) uses such signitures to classify the different items appeared in the satellite image. The classification tools used in GIS became traditional due to they used for years, which do not provide satisfactory results. Therefore, the developments are required to keep abreast of advances in technology. This paper presents a method for satellite image classification aiming at handling the problem of satellite image classification. The newly proposed method is based on two phases: Image encoding and classification based Artificial Neural Network (ANN). The first phase depends on encode the satellite image. Whereas, the second phase depends on classifying the image using the ANN, in which the input materials are an achieved codes that resulted from the first phase, while the output are the classified image regions.The classification results gave 99% classification accuracy, which is a promising score in comparison with related litritures.