Multitemporal hyperspectral satellite image analysis and classification using fast scale invariant feature transform and deep learning neural network classifier

• Published in: Earth science informatics Vol. 16; no. 1; pp. 877 - 886
• Main Authors: Vinuja, G., Devi, N. Bharatha
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.03.2023
• Subjects: Earth and Environmental Science; Earth Sciences; Earth System Sciences; Information Systems Applications (incl.Internet); Ontology; Simulation and Modeling; Space Exploration and Astronautics; Space Sciences (including Extraterrestrial Physics; Cellular Automata based Gaussian Filter; Remote sensing; Fuzzy- relevance vector machine; Fast scale invariant feature transform; Deep learning neural network classifier
• ISSN: 1865-0473; 1865-0481
• DOI: 10.1007/s12145-023-00948-2

Image classification is a frequent but still difficult subject in image processing, yet it has applications in various sectors and the medical profession, such as target tracking, object identification, and medical image processing. A Deep Learning Neural Network is used in this research to identify methods for satellite remote sensing images. The image data must be pre-processed before being applied to the Fuzzy- Relevance vector machine segmentation stage. Noise is eliminated from satellite images using a Cellular Automata-based Gaussian Filter method. The pre-processed satellite image is then segmented using the Fuzzy- Relevance vector machine Segmentation approach to achieve inverse shape identification while utilizing the least amount of energy. Following segmentation, the satellite images are subjected to Fast Scale Invariant Feature Transform feature extraction, and the Deep Learning Neural Network is utilized to classify the images. When compared to existing approaches, the proposed method’s findings have an exceptional accuracy of 98.9%.