Ex Parte Jin et al

Patent Trial and Appeal Board

Feb 16, 2017

12323210 (P.T.A.B. Feb. 16, 2017)

United States Patent and Trademark Office
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O.Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
12/323,210 11/25/2008 Hailin Jin B847 7387
108982 7590
Wolfe-SBMC
116 W. Pacific Avenue
Suite 300
Spokane, WA 99201
EXAMINER
PIERRE LOUIS, ANDRE
ART UNIT PAPER NUMBER
2123
NOTIFICATION DATE DELIVERY MODE
02/21/2017 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
docket@sbmc-law.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte HAILIN JIN and KAINI
Appeal 2016-0055931
Application 12/323,210
Technology Center 2100
Before JUSTIN BUSCH, JOHN P. PINKERTON, and ALEX S. YAP,
Administrative Patent Judges.
YAP, Administrative Patent Judge.
DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from the Examiner’s final
rejection of claims 1-22, which are all the claims pending in this application.
We have jurisdiction under 35 U.S.C. § 6(b).
We affirm.
1 According to Appellants, the real party in interest is Adobe Systems Inc.
(Br. 3.)
Appeal 2016-005593
Application 12/323,210
STATEMENT OF THE CASE
Introduction
Appellants’ invention “relate[s] to the field of consensus estimators
such as the Random Sample Consensus (RANSAC) estimator.” (Nov. 25,
2008 Specification (“Spec.”) ^ 2.) Claim 1 is representative and is
reproduced below (with minor reformatting):
1. A method comprising:
grouping, by a computing device, data points that are
derived from sets of data captured from one or more
sensors into a plurality of groups based on the sets of data
from which the data points are derived, data points derived
for one said set of data corresponding to data points in at
least one other said set of data, and corresponding data
points being determined to represent a corresponding
feature in the sets of data;
establishing a quality score for each group of the plurality
of groups that is indicative of the relative quality of the
data points in the group with respect to other said groups
and a greater number of data points in a group contributing
to a higher quality score than a fewer number of data
points in the group, groups having a higher quality score
considered more likely to include inlier data points than
groups having a lower quality score;
selecting, for a sampling process to fit a model to the data
points, one or more groups from the plurality of groups
based on the quality scores, the selected one or more
groups including a group having a highest quality score;
sampling data points from the one or more selected
groups, the sampled data points being sampled from the
one or more selected groups by a consensus estimator into
which the one or more selected groups are input to
generate a model that fits the data points; and
verifying the model across each of the data points that are
derived by inputting the data points derived into the
2
Appeal 2016-005593
Application 12/323,210
consensus estimator regardless of whether the data points
were sampled from the one or more selected groups.
Prior Art and Rejection on Appeal
The following table lists the prior art relied upon by the Examiner in
rejecting the claims on appeal:
Yokoyama et al
(“Yokoyama”)
DiFoggio et al.
(“DiFoggio”)
Adachi et al.
(“Adachi”)
US 5,602,944
US 5,668,374
US 2005/0216237 Al
Feb. 11, 1997
Sept. 16, 1997
Sept. 29, 2005
Ondrej Chum and Jin Matas, Matching with PROSAC - Progressive Sample
Consensus, CVPR, 1-7 (2005) (“Chum”).
Claims 1-22 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Adachi, in view of Yokoyama and DiFoggio, and further
in view of Chum. (See Final Office Action (mailed Apr. 10, 2015) (“Final
Act.”) 2-9.)
ANAFYSIS
We have reviewed the Examiner’s rejection in light of Appellants’
arguments that the Examiner has erred. We are not persuaded that the
Examiner erred in rejecting claims 1-22.
3
Appeal 2016-005593
Application 12/323,210
With respect to independent claim l,2 the Examiner finds Adachi
teaches or suggests “selecting, for a sampling process to fit a model to the
data points, one or more groups from the plurality of groups,” but:
Adachi do[es] not specifically state that a quality score is
established for the one or more groups, [rather it] teaches testing
hypotheses against a group of data to determine whether the
results supports the points on the surface, so as to establish a best
fit (i.e. high score or correlation) for each of the models, see para
29, 38, and further provides the uses of random consensus
algorithm for robust model fitting to accurately fit model that
describes the greatest number of local points, para 35, which
would greatly by [sic] understood by one of ordinary skilled in
the art.
(Final Act. 3-4, emphasis added.) In other words, the Examiner finds
Adachi teaches or suggests selecting one or more groups from the plurality
of groups based on quality scores, including the group with the highest
score. The Examiner further finds that Yokoyama teaches or suggests
establishing a quality score for each group:
Nonetheless, Yoko[y]ama et al. teaches using at least two sensors
13 and 14 to capture images such that group of data are made
with respect to each respective sensor (see col.3 lines 16-28), and
establishing a quality score for each group of the plurality of
groups that is indicative of the relative quality of the data points
in the group with respect to other said groups (col.3, lines 29-
36[.]
(Final Act. 4, italics in original, underline added.) Therefore, according to
the Examiner, the combination of Adachi and Yokoyama teaches or suggests
this limitation. Appellants disagree and contend that:
Although Adachi does discuss selection of groups of points,
Adachi does not select groups for performing a sampling process
2 Claims 10, 19, and 22 contain a similar limitation and the Examiner groups
these claims with claim 1 in the rejection. (Final Act. 3.)
4
Appeal 2016-005593
Application 12/323,210
to fit a model to data points that are derived from sets of data
captured from one or more sensors. Instead, Adachi describes
that one or more groups are selected to determine points that lie
on a same surface in three-dimensional (3D) space (Adachi
T|[0006]). In particular, Adachi describes that the one or more
groups are selected to draw a two-dimensional (2D) polygonal
fence around the points in those groups (Adachi f[0006]).
Selecting groups of points for determining points that lie on a
same surface in 3D space, such as by drawing a 2D fence around
those points, as described in Adachi, does not correspond to
“selecting, for a sampling process to fit a model to the data
points, one or more groups from the plurality of groups” as
recited in claim 1. This subject matter is simply missing from
Adachi. By selecting groups of related points in the manner
claimed, e.g., based on quality scores, the likelihood is higher
that the points sampled by a consensus estimator are inliers, and
thus that a reasonably accurate model may be produced.
(App. Br. 17-18, emphasis added; Reply 5-6.)
Appellants, however, have not persuaded us of Examiner error.
“[T]he Board reasonably interpreted Rule 41.37 to require more substantive
arguments in an appeal brief than a mere recitation of the claim elements and
a naked assertion that the corresponding elements were not found in the
prior art.” In reLovin, 652 F.3d 1349, 1357 (Fed. Cir. 2011); In re Geisler,
116 F.3d 1465, 1470 (Fed. Cir. 1997); In re De Blauwe, 736 F.2d 699, 705
(Fed. Cir. 1984). Here, Appellants merely state what Adachi purportedly
teaches (“Adachi describes . . .”) and conclude that this teaching does not
correspond to the limitation at issue (“Selecting groups of points for
determining . . .”). (App. Br. 21.) In addition, we agree with the Examiner’s
finding that Adachi, in view of Yokoyama, teaches or suggests this
limitation. (Final Act. 3—4; Ans. 2.) Specifically, we agree with the
Examiner’s finding that Yokoyama teaches or suggests establishing a quality
score for each group while Adachi teaches or suggests selecting one or more
5
Appeal 2016-005593
Application 12/323,210
groups from the plurality of groups based on the quality score, including a
group having a highest quality score. (Final Act. 3—4; Ans. 2.)
Appellants next contend that “Adachi fails to provide a basis for
‘sampling data points from the one or more selected groups’ as recited in
claim 1, in part, because Adachi does not include groups that are selected in
a same manner as claim IT (App. Br. 18, emphasis added.) Because of
this, Appellants contend that:
the “fit procedure” described in Adachi is not performed by a
consensus estimator relative to groups of data points. Rather,
Adachi at ^[0007] describes that a fit procedure is utilized to
generate a geographic primitive of a surface of interest from the
selected surface points, e.g., the points that were selected by
drawing the polygonal fence around the points. Thus, Adachi’s
fit procedure is performed relative to an entire collection of
individual points. It is not group-based.
(App. Br. 18-19, emphasis added; Reply 6-7.)
Appellants have not persuaded us that the Examiner erred. We agree
with, and adopt as our own, the Examiner’s findings regarding this
limitation. (Ans. 2-3, Final Act. 3-6.). As discussed above, Adachi in view
of Yokoyama teaches or suggests selecting one or more groups from the
plurality of groups based on the quality score, including a group having a
highest quality score. Sampling is then performed on the selected one or
more groups. Therefore, we agree with the Examiner that it is the
combination of the prior art references that is relied upon to reject this
limitation. See In re Keller, 642 F.2d 413, 426 (CCPA 1981) (“One cannot
show non-obviousness by attacking references individually where, as here,
the rejection is based on combinations of references.”).
Appellants further contend that because “Chum does not sample
points from selected groups ofpoints to generate a model[], the solution
6
Appeal 2016-005593
Application 12/323,210
verified by Chum does not correspond to the model verified in the
‘verifying’” limitation. (App. Br. 19, emphasis added.) The Examiner
points out, and we agree, that “Chum was not cited for the sampling step of
the claims.” (Ans. 3.) Instead, Chum was cited for “verifying the model
across each of the data points that are derived by inputting the data points
derived into the consensus estimator regardless of whether the data points
were sampled from the one or more selected groups.” (Final Act. 6; Ans. 3-
4; Chum § 2 (“The samples, unlike in RANSAC, are not drawn form [sic] all
data, but from a subset of the data with the highest quality. . . . The
hypotheses are verified against all data.”).) Appellants do not respond to the
Examiner’s explanation. (Reply 7.) Therefore, Appellants have not
persuaded us of Examiner error.
The Examiner finds that it would have been obvious to one of
ordinary skill in the art to combine the references because:
Adachi, Yokoyama, D[i]Foggio, and Chum et al. are analogous
art because they are from the same field of endeavor and that the
model analyze[d] by Chum et al. is similar to that of Adachi,
Yokoyama, and D[i]Foggio. Therefore, it would have been
obvious to one of ordinary skilled in the art to combine the
method of Chum et al. with that of Adachi, Yokoyama and
D[i]Foggio for the purpose of verifying the model and
generating] samples in an efficient manner (see page 3 lower
right column [in Chum]).
(Final Act. 6, emphasis added.) Appellants contend that the Examiner’s
“motivation to combine the references . . . lacks in particularity, and is so
sweepingly broad that it could be used to argue for obviousness of any
invention [and] rests on hindsight reconstruction.” (App. Br. 19-20.)
We agree with the Examiner that “Adachi, Yokoyama, DiFoggio, and
Chum are analogous art because they are from the same field of endeavor
7
Appeal 2016-005593
Application 12/323,210
‘3D surface points sampling and filtering from captured images’” (Ans. 4).
SeeKSR Int’l Co., v. Teleflex, Inc., 550 U.S. 398 (2007) (“The combination
of familiar elements according to known methods is likely to be obvious
when it does no more than yield predictable results.”). Here, the Examiner
has set forth “‘an articulated reasoning with some rational underpinning to
support the legal conclusion of obviousness,’” and Appellants have not
persuaded us that the Examiner erred. KSR, 550 U.S. at 415, 418.
For the foregoing reasons, we are not persuaded of Examiner error in
the rejection of independent claims 1, 10, 19, and 22. (See App. Br. 20-31
(repeating same arguments for claims 10, 19, and 22).) Thus, we sustain the
35 U.S.C. § 103 rejection of these claims, as well as the 35 U.S.C. § 103
rejection of claims 2-9, 11-18, 20, and 21, which depend on either claim 1,
10, or 19, and are not argued separately. (App. Br. 20, 24, 27-28.)
DECISION
The decision of the Examiner to reject claims 1-22 is affirmed.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l)(iv).
AFFIRMED
8