1974 年 3 巻 6 号 p. 373-382
An economical method is proposed for computing optical transfer functions with relatively small computers. The method is based on the practical view that the transfer functions in the radial and tangential directions provide the necessary and sufficient information for the evaluation of usual lenses, and by the use of the symmetry of the pupil function the requirement of large amounts of computer memories is alleviated and still the faster calculation is achieved. The validity of the method is confirmed by performing test calculations and comparing the required time and memories with those spent in the same calculations by the conventional double-transform method. Theoretical considerations are given to the maximum aberration for which the transfer function can be computed with sufficient accuracy, and the characteristics of errors introduced by the aberrations exceeding this criterion are also investigated. The theory is verified by numerical experiments.