This research aims to develop an automated system for detecting built-up areas in rural regions, addressing challenges in accurately identifying residential buildings, where the distinction between building surfaces, streets, and unpaved roads, as they exhibit similar brightness levels, is the main obstacle. In this study, we utilized high-resolution RGB satellite images from the World View-2 satellite, which has a spatial resolution of 0.46 m. We developed an automated system for detecting built-up areas. The system extracts 13 discriminative features categorized into pixel-based and area-based metrics to enhance the differentiation between non-built-up buildings and regions. These features include colour-based metrics, such as the average grey value, greyness1, and greyness2, which capture variations between colour channels, as well as hue calculations derived from advanced equations to improve surface distinction. Additionally, local statistical metrics contributed to refining detection accuracy by reducing noise. These features were selected based on their effectiveness in handling the variability in building materials and textures commonly found in rural environments. Furthermore, an Artificial Neural Network (ANN) classifier was trained to distinguish between built-up and non-built-up areas. To further enhance detection accuracy, post-processing techniques, including median filtering, gap removal, and eliminating irrelevant narrow slices, were incorporated. The system's performance was evaluated using accuracy, precision, recall, and F1-score metrics to ensure the robustness and reliability of the building detection process. The results demonstrated that the ANN classifier effectively detects buildings while overcoming challenges posed by irregular shapes and varied surface characteristics.