Python Image Processing Cookbook

There's more...

Compare the speed (run time) and quality of the outputs produced by the two algorithms. You can do image inpainting with convex optimization too (for example, the image reconstruction problem solved by the forward-backward splitting algorithm). The convex optimization problem is the sum of a data fidelity term and a regularization term, which expresses a prior on the smoothness of the solution, given by the following:

Here, ∥⋅∥_TV denotes the total variation, y are the measurements, g is a masking operator, and τ expresses the trade-off between the two terms (a regularization parameter).