Abstract
This presentation will focus on segmentation and reconstruction problems using an integer linear programming approach. These problems have important applications in remote sensing, medical image analysis and industrial inspection. We focus on methods that produce optimal or near-optimal solutions for the corresponding optimization problems. We show that for the two problems one may use similar ideas in both modeling and solution methods. These methods are based on Lagrangian decomposition and dynamic programming for certain subproblems (associated with lines in the image).
