Appeal No. 2006-1618
Application No. 10/046,797
The examiner responds that Catros uses gradient amplitudes in an
algorithm to determine the best path to connect a disjointed contour [answer,
page 14]. According to the examiner, since these amplitudes are numerical
values associated with a gradient, they reasonably teach weight values [id.]. The
examiner further notes that Makram-Ebeid teaches eliminating interfaces when a
scale parameter increases. Therefore, each eliminated interface has an
associated scale parameter at which it exists [id.]. Moreover, the examiner
indicates that when two regions are merged in Makram-Ebeid, the function
"Energy" is minimized thus likely minimizing intensity variance [answer, pages 14
and 15].
We will sustain the examiner's rejection of claim 20. Makram-Ebeid
discloses an algorithm that eliminates the largest possible number of interfaces
to ultimately merge adjacent regions with practically identical intensities
[Makram-Ebeid, col. 1, lines 43-45]. To this end, the algorithm uses the Energy
function that comprises (1) a first term that accounts for intensity variance in each
image region, and (2) a second term that accounts for the total length of the
image's boundaries, weighted by a scale parameter (λ) [Makram-Ebeid, col. 1,
lines 48-53]. Initially, the value "1" is assigned to the scale parameter λ and
adjacent regions are merged that minimize the Energy function. The resultant
regions are then re-organized and, following a recalculation of the Energy
function terms, a new merger attempt is made with λ=1. This operation is
repeated until there is no longer any region with an adjacent region to merge
21
Page: Previous 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Next
Last modified: November 3, 2007