Polarization-sensitive optical image measurement is subject to a non-negligible bias, and consequent deviation in birefringence, in a surrounding range of low SN ratios (signal-to-noise ratios) and low signal strengths however, this deviation in birefringence is removed to make accurate quantitative measurement possible. Noise-containing OCT signals obtained by polarization OCT are processed using a birefringence calculation algorithm, to obtain measured birefringence, after which noise is statistically adjusted to simulate a measured birefringence distribution and determine the noise characteristics of the measured birefringence values, and then Monte Carlo calculations are repeated by assuming different values for the noise level and the true birefringence value, respectively, to form three-dimensional histogram of combinations of true birefringence values, SN ratios, and measured birefringence values, after which specified measured birefringence values and SN ratios are assumed from the three-dimensional histogram information to obtain a true birefringence probability density distribution, and true birefringence values are estimated from the true birefringence probability density distribution.
