We propose a novel phase-shift calibration algorithm. With this technique w
e determine the unknown phase shift, between two interferograms by examinin
g the sums and differences of the intensities on each interferogram at the
same spatial location, i.e., I-1(x, y) +/- I-2(x, y) These intensities are
normalized so that they become sinusoidal in form. A uniformly illuminated
region of the interferograms that contains at least a 2 pi variation in pha
se is examined. The extrema of these sums and differences are found in this
region and are used to find the unknown phase shift. An error analysis of
the algorithm is provided. In addition, an error-correction algorithm is im
plemented. The method is tested by numerical simulation and implemented exp
erimentally. The numerical tests, including digitization error, indicate th
at the phase step has a root-mean-square (RMS) phase error of less than 10(
-6) deg. Even in the presence of added intensity noise (5% amplitude) the R
MS error does not exceed 1 deg. The accuracy of the technique is not sensit
ive to nonlinearity in the interferogram. (C) 2000 Optical Society of Ameri
ca [S0740-3232(00)01711-7].