Summary: Human Vision, Visual Processing, and Digital Display, eds. B.E. Rogowitz and J. Allebach, Proc. Vol.
1249, 124-134, SPIE, Bellingham, WA.
Learning receptor positions from imperfectly known motions
Albert J. Ahumada, Jr. and Jeffrey B. Mulligan
NASA Ames Research Center, Human Interface Branch
MS 239-3, Moffett Field, California, 94035
An algorithm is described for learning image interpolation functions for sensor arrays whose sensor
positions are somewhat disordered. The learning is based on failures of translation invariance, so it
does not require knowledge of the images being presented to the visual system. Previously reported
implementations of the method assumed the visual system to have precise knowledge of the
translations. We demonstrate here that translation estimates computed from the
imperfectly interpolated images can have enough accuracy to allow the learning process to converge to
a correct interpolation.
The human visual system has the capability of making fine geometric estimates even though the
receptor array is disordered,1
variable in density,2
and changes after birth as the eye grows and the
fovea becomes tightly packed.3