Goksel Dedeoglu et al.

Patent Trials and Appeals Board

Jul 30, 2019

13270365 - (D) (P.T.A.B. Jul. 30, 2019)

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.
13/270,365 10/11/2011 Goksel Dedeoglu TI-70112 1038
23494 7590 07/30/2019
TEXAS INSTRUMENTS INCORPORATED
P O BOX 655474, M/S 3999
DALLAS, TX 75265
EXAMINER
LUO, KATE H
ART UNIT PAPER NUMBER
2488
NOTIFICATION DATE DELIVERY MODE
07/30/2019 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):
uspto@ti.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
____________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________________

Ex parte GOKSEL DEDEOGLU and VINAY SHARMA
____________________

Appeal 2018-008716
Application 13/270,3651
Technology Center 2400
____________________

Before ALLEN R. MACDONALD, STACY B. MARGOLIES, and
IFTIKHAR AHMED, Administrative Patent Judges.

AHMED, Administrative Patent Judge.

DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from a final rejection of
claims 2, 5–9, 11, and 12, which are all of the claims pending in the
application. Claims 1, 3, 4, and 10 have been cancelled. App. Br. 2; Dec. 1,
2017 Advisory Action (entering amendment cancelling claims 1, 3, 4, and
10). We have jurisdiction under 35 U.S.C. § 6(b).
We REVERSE.
SUMMARY OF THE INVENTION
The invention generally relates to a “depth-fill algorithm for low-
complexity stereo vision.” Spec. ¶ 2.

1 According to Appellants, the real party in interest is Texas Instruments
Incorporated. App. Br. 2.
Appeal 2018-008716
Application 13/270,365

2
Claims 5 and 9 are illustrative of the subject matter on appeal and are
reproduced below with certain limitations at issue emphasized:

5. An apparatus for low-complexity stereo vision, comprising:
a memory;
a processor coupled to the memory and configured to
obtain a depth image based on left and right images of a
stereo camera by tracking a farthest depth value on a pixel
basis for each of a plurality of pixels over a plurality of
stereo frames.

9. A non-transitory computer readable medium comprising
instructions that, when executed by a processor, cause the
processor to:
estimate a first depth value for a pixel in a depth model
based on two or more previously-processed stereo frames,
each of the previously-processed stereo frames including
respective left and right images obtained from a stereo
camera;
estimate a second depth value for the pixel based on left
and right images of a currently-processed stereo frame
obtained from the stereo camera;
determine whether the second depth value is deeper than
the first depth value in the depth model; and
update the first depth value in the depth model with the
second depth value in response to determining that the
second depth value for the pixel is deeper than the first
depth value for the pixel in the depth model.
REJECTION
Claims 2, 5–9, 11, and 12 stand rejected under 35 U.S.C. § 103(a) as
obvious over the combination of Izzat et al. (US 2009/0167843 A1; publ.
Appeal 2018-008716
Application 13/270,365

3
July 2, 2009) and Hughes et al. (US 2009/0247249 A1; publ. Oct. 1, 2009).
Final Act. 3.
ISSUES
1. Did the Examiner err in concluding that Izzat teaches or suggests
“estimat[ing] a first depth value for a pixel in a depth model based
on two or more previously-processed stereo frames, each of the
previously-processed stereo frames including respective left and
right images obtained from a stereo camera,” as recited in claim 9,
and similarly recited in claims 2 and 6?
2. Did the Examiner err in rejecting claim 5 under 35 U.S.C. § 103(a)
as being obvious over the combination of Izzat and Hughes?
ANALYSIS
Claims 2, 6–9, 11, and 12
Independent claim 9 recites a non-transitory computer readable
medium comprising instructions that, when executed by a processor, cause
the process to “estimate a first depth value for a pixel in a depth model based
on two or more previously-processed stereo frames, each of the previously-
processed stereo frames including respective left and right images obtained
from a stereo camera.” App. Br. 19 (emphasis added). Independent claim 2
(a method claim) and dependent claim 6 (an apparatus claim) recite similar
limitations. Id. at 18–19.
The Examiner relies on Izzat for disclosure of this limitation, finding
that Izzat discloses that depth estimation 302 (shown in Figure 3) matches
the pixels in the left image to those in the right image and “then a
triangulation procedure 303 is used to convert the disparity values to the
Appeal 2018-008716
Application 13/270,365

4
depth values.” Final Act. 7 (citing Izzat, Fig. 3, ¶ 26). The Examiner
characterizes this disclosure in Izzat as “a two-pass technique for the
recovery of three dimension 3D information.” Ans. 12. According to the
Examiner, in Izzat, “[a] first pass in a static scan phase recovers a three
dimension of a static scene using a low speed, high accuracy technique.”
Ans. 12–13. The Examiner adds that “[s]ince static scene scanning would
need to be repeated multiple times to recover any new items introduced in
the static scene, the multiple times scanning results in multiple stereo
frames.” Id. at 13. Thus, according to the Examiner, “the first depth values
from static scene scanning are generated under multiple stereo frames.” Id.
Appellants contend that the depth values in Izzat are “calculated based
on the disparity between left and right images of a single stereo image pair”
and that “[n]o mention is made in Izzat et al. of estimating a first depth
valued based on two different stereo frames.” App. Br. 6. Appellants also
argue that the depth value in paragraph 26 of Izzat is generated as a part of a
“dynamic scan phase,” and not as part of a “static scan phase,” as Examiner
asserts. Reply Br. 2–3. According to Appellants, Izzat “does not mention
that ‘depth values’ are computed during the ‘static scan phase,’” or mention
“using ‘stereo frames’ for the ‘static scan phase.’” Id. at 3. Appellants add
that Izzat “does not discuss using ‘stereo frames’ during the ‘static scan
phase.’” Id. Instead, Appellants argue, Izzat “appears to use different non-
stereo views” during the static scan phase. Id. (citing Figs. 1, 2). Therefore,
Appellants conclude, “the Examiner’s argument that the ‘static scene
scanning’ is repeated in Izzat et al. does not change the fact that the depth
values [are] generated in the ‘dynamic scan phase’ of Izzat.” Id. at 2–3.
Appellants also argue that “even if the ‘static scene scanning’ in Izzat et al.
Appeal 2018-008716
Application 13/270,365

5
were to generate depth values . . . , such depth values would not be generated
[‘]based on two or more previously-processed stereo frames,’ as positively
recited in pending claim 9.” Id. at 3.
Ultimately, we are persuaded of Examiner error. We agree with
Appellants that the Examiner has not shown that Izzat teaches estimating a
first depth value based on two or more previously-processed frames, each of
the frames including respective left and right images. As illustrated in
Figure 3 and described in paragraph 26, Izzat discloses that “[a] stereo image
pair is subjected to multiple steps of processing,” including disparity
estimation by matching “the pixels in the left image to those in the right
images,” and a triangulation procedure that converts the disparity values to
depth values, resulting in “depth map 302 . . . obtained from the stereo
image pair.” Izzat ¶ 26. The Examiner has not shown that Izzat discloses
that the stereo image pair processing is part of the static scan phase or that
the stereo image pair processing “is repeated multiple times” to generate
depth values based on multiple previously-processed stereo frames. See
Ans. 13. Rather, we agree with Appellants (see Reply Br. 2) that Izzat
discloses that depth map 302 described in paragraph 26 (on which the
Examiner relies for the claimed “first depth value”) is generated during the
dynamic scan phase, not during the static scan phase (see Izzat ¶¶ 16, 18,
26).
Specifically, Izzat discloses a “two-pass approach [that] recovers the
geometry of the static background and dynamic foreground separately using
different methods.” Id. ¶ 18 (emphasis added). Izzat discloses that, in the
static scan phase, “a high accuracy 3D acquisition approach is used” and
that, in the dynamic scan phase, “the dynamic acquisition of 3D information
Appeal 2018-008716
Application 13/270,365

6
needs to be performed with a fast, possibly less accurate method of
scanning.” Id. ¶¶ 21, 22. The Examiner does not persuasively explain how
disparate disclosures in Izzat regarding different scan phases teach the
claimed estimating a first depth value. Izzat does not disclose that, during
the dynamic scan phase, depth map 302 (which is based on a left image and
a right image) is created based on previously-processed frames. Id. ¶ 26.
Nor does Izzat disclose that the static scan phase, which may be repeated to
recover new items introduced in the scene (id. ¶ 16, claims 6, 15), creates a
depth value that is based on previously-processed stereo frames including
left and right images.
Thus, we agree with Appellants that the Examiner has not shown that
Izzat teaches or suggests “estimat[ing] a first depth value for a pixel in a
depth model based on two or more previously-processed stereo frames, each
of the previously-processed stereo frames including respective left and right
images obtained from a stereo camera,” as recited in claims 2, 6, and 9. The
Examiner also does not rely on Hughes as teaching this claim limitation in
support of the obviousness rejection based on Izzat and Hughes.
Accordingly, we do not sustain the Examiner’s obviousness rejection of
claims 2, 6–9, 11, and 12.
Claim 5
Independent claim 5 recites “a processor coupled to the memory and
configured to obtain a depth image based on left and right images of a stereo
camera by tracking a farthest depth value on a pixel basis for each of a
plurality of pixels over a plurality of stereo frames.” App. Br. 18 (emphases
added).
Appeal 2018-008716
Application 13/270,365

7
The Examiner again relies on the static and dynamic scan phases
disclosed in Izzat as teaching “obtain[ing] a depth image based on left and
right images of a stereo camera . . . over a plurality of stereo frames.” Final
Act. 6 (citing Izzat ¶¶ 16, 26, Fig. 3). The Examiner also finds that Hughes
discloses “tracking a farthest depth value on a pixel basis over a plurality of
stereo frames” because it discloses a Z-buffer to store the depth information
for pixels and discloses “Z-buffer tests [whereby] the new pixel replaces the
previous pixel value if its depth is greater than or equal to the previous value
stored in the Z buffer.” Id. (citing Hughes ¶ 40, Fig. 1).
Appellants argue that there is no mention “of tracking a farthest depth
value over a plurality of different stereo frames” in either Izzat or Hughes.
App. Br. 11–12. According to Appellants, the depth values in Izzat “are
calculated based on the disparity between left and right images of a single
stereo image pair,” not over a plurality of stereo frames. Id. at 11. Hughes,
Appellants contend, discloses that “a single image or frame may be
composed of a large number of polygons that are separately and sequentially
rendered” and “Z-buffering is performed for each polygon that is rendered in
a frame.” Id. at 12 (citing Hughes ¶ 40). Appellants argue “Z-buffering is
performed [in Hughes] by comparing depth values that are generated for
different primitives or polygons of the same graphics frame.” Id.
Therefore, Appellants argue, Hughes fails to teach or suggest tracking a
depth value over a plurality of different stereo frames. Id.
We are persuaded that the Examiner has erred. As discussed above,
the Examiner has not shown that Izzat discloses that the stereo image pair
processing (that calculates depth values) is part of the static scan phase or
that the stereo image pair processing “is repeated multiple times” to generate
Appeal 2018-008716
Application 13/270,365

8
depth values based on multiple stereo frames. See Final Act. 6. Rather, we
agree with Appellants (see App. Br. 11) that Izzat discloses that depth map
302 is generated during the dynamic scan phase, not during the static scan
phase, and is based on left and right images of a single stereo image pair (see
Izzat ¶¶ 16, 18, 26). Claim 5, however, requires “obtain[ing] a depth image
based on left and right images . . . by tracking a farthest depth value . . . over
a plurality of stereo frames.” Emphasis added.
To the extent the Examiner relies on Hughes for teaching this feature,
we also agree with the Appellants (see App. Br. 12) that Hughes discloses
that depth tracking is performed by comparing depth values that are
generated for different primitives or polygons, and there is no express
disclosure that this comparison involves comparing pixel values from a
plurality of stereo frames (see Hughes ¶ 40). Hughes discloses a “2D to 3D
texture mapping process to add visual detail to 3D geometry.” Id. Hughes
further discloses that an image is “constructed from basic building blocks
known as graphics primitives or polygons,” and the texture mapping process
performs depth tracking over multiple polygons of the same image in
rendering that image to a display. Id. The Examiner does not persuasively
explain how this disclosure in Hughes regarding depth tracking over a
plurality of polygons teaches the claimed depth tracking over “a plurality of
stereo frames.”
Thus, we agree with Appellants that the Examiner has not shown that
the combination of Izzat and Hughes teaches or suggests “obtain[ing] a
depth image based on left and right images of a stereo camera by tracking a
farthest depth value on a pixel basis for each of a plurality of pixels over a
Appeal 2018-008716
Application 13/270,365

9
plurality of stereo frames,” as recited in claim 5. Accordingly, we do not
sustain the Examiner’s obviousness rejection of claim 5.
DECISION
For the reasons above, we reverse the Examiner’s decision rejecting
claims 2, 5–9, 11, and 12.
REVERSED