Conclusions

2D SMM implementation is simple - just replace the (1,-2,1) Laplacian coefficients in the differencing star with precomputed sets of geometry-dependent coefficients. It is also numerically stable. The images produced are clear and crisp, the only artifacts being the spurious reflections off the abrupt variations in grid size. There are several possible ways of eliminating them. These artifacts are not likely to be generated in two particular types of practical applications - reflection measurements made from a platform moving with nonuniform speed or on a slightly-perturbed regular grid. Extension to 3D acquisition geometries is not impeded by any obvious obstacle.