CLASSIFICATION OF SATELLITE IMAGES WITH DEEP CONVOLUTIONAL NEURAL NETWORKS AND ITS EFFECT ON ARCHITECTURE

This study explores the application of Deep Convolutional Neural Networks (DCNNs) for the classification of satellite images and investigates the impact of various architectural considerations on model performance. The process begins with preprocessing the input satellite images using fusion techniques, including image resizing, reduction of atmospheric haze, histogram equalization, and enhancement of low-light images. The deep learning phase involves the implementation of a DCNN, a powerful neural network architecture particularly adept at capturing spatial dependencies in image data. The primary objective is image classification, categorizing scenes into distinct classes such as Cloudy, Desert, Green Area, and Water. The evaluation metrics employed to assess the model's performance include accuracy, precision, recall, and F1 score, providing a comprehensive analysis of its ability to correctly classify images. By systematically exploring the impact of architectural choices, preprocessing techniques, and performance metrics, this research contributes insights into optimizing the classification of satellite images, which has significant implications for applications ranging from environmental monitoring to urban planning and disaster management. The findings underscore the importance of tailored deep learning architectures and preprocessing strategies in achieving accurate and reliable satellite image classification results.

Keywords: Pre-Processing, Convolutional Neural Networks, Deep learning, Enhancement, Classification, Accuracy.