Abstract

In a recent paper Frieden¹ unified the theory behind many nonlinear image restoration methods. He proposed that these methods be implemented using the Newton-Raphson iterative approach. Although this approach is fast, exhibiting quadratic convergence, it requires complex programming and large core storage capabilities even for modest image sizes, unless the problem can be reduced to line-by-line processing². By contrast, recursive or replacement formulas permit interactive processing with modest storage requirements, albeit with slower convergence. They can also be implemented on a specialized image processing computer since the formulas use only simple algebraic operations. These recursive algorithms also have the capability of immediate display of intermediate images, which is especially useful in interactive processing.

© 1986 Optical Society of America