Abstract
Satellite images are vital nowadays, and it is used in remote sensing, weather forecasting, monitor day to day national security issues, etc. Due to large capacity, resolution of the earth imaging sensors and image acquisition techniques, image storage, and retrieval of images are becoming a problem nowadays. Images are retrieved using text-based image retrieval that was unable to provide accurate results. Content-based image retrieval (CBIR) for remote information system provides the solution for the retrieval problem. Existing CBIR system based on the low-level feature extraction techniques such as color, shape, and texture remains an open problem. In this research, we proposed and implemented a new fusion technique for extracting the features of the images, based on the hybrid model of the extraction, which includes sparse auto-encoding of gray-level co-occurrence matrix-based features from principal component analysis and discrete wavelet transform derived coefficients, with prediction of information using neural network by applying the fuzzy inference system approach for the classification of the retrieved data. Results obtained from the proposed technique are compared with the existing techniques gray-level co-occurrence matrix, scale invariant feature transform, and speeded up robust features. The experimental results show that the performance of the proposed system is better in terms of accuracy, precision, and error in comparison with previous techniques.