Appeal No. 1999-2069
Application 08/397,639
line 29). Although Smilansky discusses rigid transformations
(e.g., col. 11, lines 42-43), which involve rotation and
translation, but not scaling, equation 1b shows the use of
scaling in the general case, where S x and Sy are scaling factors
(and where " would normally be B/2 for no angular deviation of
the sensor). Thus, Smilansky discloses affine transformations
involving rotation, translation, and scaling (an enlargement or
reduction factor). Smilansky discloses that a full affine
transformation can be computed based on the theory of least-
squares data fitting (col. 11, lines 15-34).
Frankot discloses that "[f]or any transformation more
general than pure translation, scale factor (magnification) or
rotation for example, registration effectiveness depends on the
relative positions of each measurement in addition to the
accuracy of the measurements themselves" (col. 1, lines 59-63),
where we note that translation, rotation, and scale factor are
affine transformations. Frankot discloses selection criteria for
automatic subarea selection to improve image registration.
Frankot discloses that "[i]mage registration requires estimation
of the coordinate transformation f that aligns two images"
(col. 6, lines 4-6). Frankot discloses that the transformation f
may be an affine transformation (col. 8, lines 65-67) and that
the coordinate transformation may be fitted with a weighted-
least-squares method (col. 6, lines 23-30).
- 10 -
Page: Previous 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Next
Last modified: November 3, 2007