Study of Snake Active Contour Model In Building Extraction from Quickbird Image

• Main Author: FARIDA WIDYANINGRUM (NIM 25106014), BERNADETA
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/10097
• Institution: Institut Teknologi Bandung

Development of high resolution satellite imagery simplifies object identification such as building, road and other nature objects without direct contact with the objects. Feature extraction is one of methods to obtain information from satellite imagery. Extraction can be defined as a process to gather information needed from image. Snake active contour model is a well known method in a computer vision, utilized for structured and unstructured object extraction from digital imagery. This research intended to apply snake active contour model to extract building from Quickbird imagery. This research modified snake active contour model in order to extract multi-object, in this case, building from Quickbird imagery. Some experiments had been conducted to facilitate the influence of initialization and weight of energy function to obtain appropriate algorithm that offer optimal result. Quality assessment of building extraction is conducted by comparing it with reference data (on screen digitized data). Quality assessment consisted of 100% Overall Quality, 100% correctness, 100% completeness, 0.1276 m2 standard deviation of similarity quality and 2.527 m standard deviation of location quality.