Section outline

    • Premise. This course provides an introduction to computational imaging, from mathematical/physical models of image formation to modern reconstruction methods combining model-based optimization and data-driven learning. We will discuss ill-posedness, variational and Bayesian regularization, first-order optimization algorithms, and learning-based approaches such as algorithm unrolling. Particular attention will be given to how physical models, mathematical structure, and learned components can be integrated to obtain accurate, efficient, and interpretable imaging methods, with examples and hands-on notebooks illustrating the theory in practice.
    • Aims. The course aims to provide the mathematical and computational foundations for formulating and solving imaging inverse problems using model-based, learning-based, and hybrid approaches.
    • Exam. Oral exam consisting of a discussion of a computational project previously agreed with the teacher and questions on the material covered in the course.
    • Student assistance hours: please contact the instructor (luca.calatroni@unige.it).