BLIND DECONVOLUTION OF A HERMITIAN AND A NON-HERMITIAN FUNCTION

Blind deconvolution is the problem of recovering two functions from their convolution. We treat the blind-deconvolution problem under restricted conditions that the components of the convolution are Hermitian and non-Hermitian functions and that the support of the non-Hermitian function is known. This problem is solved by combining a method for retrieving the Fourier phase of the non-Hermitian function from a convolution with a phase-only reconstruction algorithm. The characteristic of the combined method is that the uniqueness property of its solution is understood from the theory of analytic functions. A number of results obtained from computational implementation are also presented.