**Computer vision**

**Werkhoven P., Toet A. &
Koenderink J.J. (1990). Displacement estimates through adaptive affinities.
IEEE Transactions on Pattern Analysis and Machine Intelligence 12(7),
658‑663. **Algorithms for the computation of optic flow from image
sequences by means of "local constraint" (relating spatial to temporal gray
value variations in the image plane) often postulate an additional "oriented
smoothness constraint" (assuming that the flow vector field varies smoothly).
Oriented smoothnes in addition to "local constraint" proves to be a mathematical
foundation for many iterative algorithms that compute optic flow. In this
correspondence we present a new technique that is functionally equivalent to the
oriented smoothness concept and reduces numerical complexity and computational
costs by eliminating the smoothness requirement from the iteration process. We
applied local affine transformations for propagating uniquely computed flow
vectors into homogeneous regions and along edges, in a single step. The window
within which the local affine transformation is performed, is adaptive to the
local structure of the intensity pattern, which is accordance with the oriented
smoothness concept as formulated by Nagel.

** Werkhoven P. &
Koenderink J.J. (1990). Extraction of motion parallax structure in the visual
system I. Biological Cybernetics 63, 185‑191. **We
present a paradigm to estimate local affine motion parallax structure from a
varying image irradiance pattern. The method is based on a matching of jets of
irradiance rather than the local image irradiance pattern or features of it. It
does not put any constraints on the structure of the image irradiance pattern.
Moreover, the aperture problem does not arise and additional ad hoc constraints
on the velocity field like the 'smoothness constraint'' are superfluous. An
implementation is designed such that the affine structure of the local motion
parallax field is represented through simple combinations of the outputs of
physiologically plausible receptive fields and their temporal derivatives.

**Werkhoven P. &
Koenderink J.J. (1990). Extraction of motion parallax structure in the visual
system II. Biological Cybernetics 63, 193‑199. **
We
have developed an algorithm to extract local affine motion parallax structure of
a varying image irradiance pattern in part I. In part II, we present
computational results of this algorithm for situations where the information
about the affine structure is contained in only two successive images. This
applies to a large class of problems (e.g. two-image motion sequences and
stereoscopic vision). The results show the effects of spatio-temporal
discretisation and internal noise on the performance of this algorithm.