Abstract

The challenge in modeling spatial vision is to predict resolution acuity and separation acuity (hyperacuity) in a variety of different conditions. Our viewprint model¹ successfully predicts thresholds for a high-contrast three-line bisection task. We now apply the viewprint model to a variety of two-multipole separation tasks. Several parameters were varied: contrast; stimulus blur; multipole type (ramp, edge, line, dipole, and quadrupole); polarity (opposite polarity edges, for example, correspond to a bar); and viewprint blur (to simulate sparse sampling). Three regimes of separation are found, each with different dependence on the above parameters. The resolution regime, for the smallest separations, is quite sensitive to most of the parameters varied. The null point regime between 2 and 10 min is relatively independent of parameter manipulations. The large separation regime requires modification of the model to take into account the cortical sampling grain. Our predictions will be compared to psychophysical separation thresholds for the same and opposite polarity lines in several blur and contrast conditions. To a first approximation there is good agreement between theory and experiment including some counterintuitive results.

© 1987 Optical Society of America