Comparison of two nonlinear constrained algorithms for 3D image restoration

The results from applying two constrained nonlinear image restoration algorithms to 3D optical microscopy data are presented. Both algorithms are iterative and use a priori knowledge to impose constraints on the solutions. The first algorithm uses the positivity constraint, while the second algorithm is a combination of least-squares and the method of projection onto convex sets (POCS). The positivity and a bound on the noise level are incorporated as constraints in the latter algorithm. Both algorithms give similar results but require different numbers of iterations, the latter converging much faster. Details of computation time and convergence properties are given, along with typical images processes by both algorithms for comparison