Browse Prior Art Database

Algorithm for the Automated Registration of Parallel-Projected Images

IP.com Disclosure Number: IPCOM000111536D
Original Publication Date: 1994-Mar-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 59K

Publishing Venue

IBM

Related People

Pritt, MD: AUTHOR

Abstract

Disclosed is an algorithm for the automated registration of parallel-projected images. The innovative feature of the algorithm is the use of an "epipolar constraint" to constrain the search for image-matching points, thus reducing registration error.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Algorithm for the Automated Registration of Parallel-Projected Images

      Disclosed is an algorithm for the automated registration of
parallel-projected images.  The innovative feature of the algorithm
is the use of an "epipolar constraint"  to constrain the search for
image-matching points, thus reducing registration error.

      Image registration is the process of geometrically aligning two
images of the same scene.  A registration function assigns to each
point (x, y) in the first image its corresponding point F(x, y) in
the second image.  The registration function for parallel-projected
images (i.e., images which are produced by parallel rather than
perspective projections) has the form "F"(x, y) =  A(x, y) + h(x,
y)"e", where A(x, y) is an affine transformation, h(x, y) is a
function that depends on the topographic heights, and "e" is a fixed
vector [*].

      The algorithm identifies matching points (i.e., pairs of image
points which correspond to the same point in the scene) using
two-dimensional cross-correlation methods [*], although other methods
of match point identification could be used instead.  The match
points are used to estimate the parameters of the registration
function.  The search for additional match points is then a
one-dimensional search along the lines defined by the vector "e".
This so-called "epipolar constraint"  greatly increases the speed and
accuracy of the registration.  The step-by-step algorithm is as
follows:

Step 1.  A set of match points is identified using two-d...