Abstract: Topological denoising means minimizing the number of critical points within a certain tolerance from a given input function. We consider methods based on homological persistence and TV/L2-minimization and discuss how features of the two approaches can be combined.