A visual and numerical comparative study of some parallel affine projection algorithms for solving the convex feasibility problem with application to scratch inpainting

The paper compares four variants of algorithms that solve the problem of Convex Feasibility using affine combinations of projections, two classical variants of Parallel Projection Method (PPM) and two modified variants that involve variable weight, in terms of their effectiveness in inpainting a convex polygon, as well as in terms of their convergence in a finite a number of step. We also present a numerical study of the dependence of the efficiency and the execution speed of these algorithms on the shape of the inpainted convex set, as well as on the values of the relaxation parameter.