An inverse problem refers to the inference of latent model parameters from a set of noisy measurements under a generative model that explains the causal relationship between both. Despite the broad field of applications, this typically leads to optimization problems that can be tackled by common mathematical tools. In computer vision and image processing, many tasks of practical relevance can be formulated as inverse problems ranging from low-level vision to image analysis and scene understanding. This seminar focuses on theory along with some of the classical applications of inverse problems in image processing and computer vision. The topics covered by the seminar include but are not limited to: - image filtering and denoising - blind image restoration, upsampling and super-resolution - motion estimation - inpainting - image segmentation - shape-from-X and 3-D reconstruction

Dates & Rooms:
Monday, 8:00 - 10:00; Room: KH 1.021