Uncertainty in Image-Based Change Detection

Digital aerial imagery provides the perfect medium for capturing geospatial information. The automation of object extraction from digital imagery has received substantial attention over the last few years in the computer vision and digital image analysis communities, resulting in significant progress in algorithms and theoretical concepts. In this paper we introduce a novel approach to object extraction from digital imagery using accuracy information. We extend the model of deformable contour models (snakes) to function in a differential mode. Prior information with accompanying accuracy estimates is used as input in our methodology. In a departure from standard techniques, the objective of our object extraction is not to extract yet another version of an object from the new image, but instead to update previously collected GIS information. By incorporating accuracy information in our technique we identify local or global changes to this prior information, and update the corresponding GIS information accordingly.