shown in Figure 3.2. Each aperture is positioned to cause a quarter-
wave phase shift by increased path length (detour phase). The two
non-negative quadrature terms are weighted to vector sum to the
appropriate magnitude and phase for each pixel. The two appropriate
apertures are opened according to their weight. The Lee method uses
continuous-tone variables to represent the two non-zero components.
The phase is no longer quantized by the location of the aperture. The
phase is determined by the vector addition of the two non-zero
components. In a totally binary application of the Lee method, the
apertures are rectangles positioned to obtain the quarter-wave shift.
The area of each aperture is adjusted to determine the amplitude of
each component. Once again, in this binary case, the dynamic range is
limited by the number of resolution elements within one cell.
Burckhardt10 showed that while the Lee method decomposes the
complex-valued F(u,v) into four real and positive components, only
three components are required. Each cell can be represented by three
components 1200 apart. Any point on the complex plane can be
represented as a sum of any two of these three components. As in the
Lee method, two non-negative components are chosen to represent each
cell. Because only three instead of four components have to be
stored, the required memory size and plotter resolution are reduced.
Haskell11 describes a technique in which the hologram cell is divided
into N components equally spaced around the complex plane. It is
identical to the binary Lee (N=4) and the Burckhardt (N=3) where N may
take larger values. This Generalized Binary Computer-Generated
Hologram (GBCGH) uses N columns and K rows of subcells. Each subcell
can take a transmittance value of 1 or 0. The phase is delayed by 2/N