We have studied the problem of the PCH in the GI2T data processing. This
PCH appearing in the AC of fringe pattern is produced by the CP40 detector
itself which can detect only 1 photon per pixel and per exposure time of
20 ms. We propose to remove the corresponding artifact in the power
spectrum by fitting a 2D polynomial function to the Fourier transform of
the PCH in order to reconstruct the original unbiased power spectrum. Among
the a posteriori correction methods, it is certainly the easier and
the most efficient one. Actually, the Hoffman method (Hoffman 1993)
requires the PCH calibration on a laboratory source and the method, which
consists in interpolating the PCH in the direct space, is less efficient
because the hole has a narrow support at the center of the AC. Besides, in
terms of the signal to noise ratio, the present a posteriori
correction method is superior to CC techniques (Thiébaut 1994). We have
also shown a significant improvement in visibility measurements due to the
correction over previous GI2T data reduction (M94b). After the PCH
removal, the visibility estimate becomes signal to noise independent and
more reliable. This has been demonstrated both on simulated and actual data from GI2T observations of 
Cep and PCyg.
However, all these improvements are strongly dependent on the fit accuracy. The PCH removal will be correct
only if the signal support in the spectral density masks only a small part of the PCH Fourier
transform. Consequently, the fringes must be oversampled with respect to the PCH size instead of the detector
pixel. Therefore, provided a correct oversampling, this method can be applied to other imaging techniques.
Acknowledgements
We are grateful to E. Thiebaut and R. Foy for many useful discussions and to P. Stee for a critical review of the original manuscript. The GI2T is supported by the PNHRAA french organisation from CNRS-INSU.