In this paper a special class of nonlinear Fredholm integral equations
of the first kind, the so-called Urysohn equation, is considered, whe
re the kernel depends on t only via the unknown function x(t). To over
come the ambiguity, a decreasing rearrangement approach is used. Moreo
ver, a constrained least squares method helps regularizing the problem
. As a specific property, the equation can be decomposed into a well-p
osed nonlinear part, the inversion of a function, and an ill-posed lin
ear part, a linear Fredholm integral equation of the first kind. The l
inear part of this two-stage procedure was already discussed in [8]. I
n the present paper the two-stage procedure is compared with a one-sta
ge nonlinear least squares approximation which is directly applied to
the nonlinear original integral equation. The comparison is explained
by means of a computational case study for a specific example arising
in optics.