TRW Automotive U.S.LLCv.Magna Electronics Inc.

Patent Trial and Appeal Board

Jun 25, 2015

12640425 (P.T.A.B. Jun. 25, 2015)

Trials@uspto.gov Paper 39
Tel: 571-272-7822 Entered: June 25, 2015

UNITED STATES PATENT AND TRADEMARK OFFICE
_______________
BEFORE THE PATENT TRIAL AND APPEAL BOARD
_______________
TRW AUTOMOTIVE US LLC,
Petitioner,
v.
MAGNA ELECTRONICS INC.,
Patent Owner.
_______________

Case IPR2014-00266
Patent 7,994,462 B2
_______________

Before JUSTIN T. ARBES, BENJAMIN D. M. WOOD, and
NEIL T. POWELL, Administrative Patent Judges.

WOOD, Administrative Patent Judge.

FINAL WRITTEN DECISION
35 U.S.C. § 318(a) and 37 C.F.R. § 42.73

IPR2014-00266
Patent 7,994,462 B2

2

I. INTRODUCTION
A. Background
TRW Automotive US LLC (“TRW”) filed a Petition (Paper 2, “Pet.”)
to institute an inter partes review of claims 30–32, 34–41, 43–46, 48, and 49
of U.S. Patent No. 7,994,462 B2 (Ex. 1002, “the ’462 patent”). Magna
Electronics Inc. (“Magna”) filed a Preliminary Response. Paper 9. In our
Decision on Institution (Paper 17, “Dec.”), we instituted an inter partes
review of claims 30, 34, and 38 based on the proposed ground that these
claims were unpatentable under 35 U.S.C. § 102(b) as anticipated by
Kenue.1 Dec. 16. We subsequently granted-in-part TRW’s Request for
Rehearing and instituted review of claims 31 and 37 on the same proposed
ground of unpatentability. Paper 21, 4–5.
After the Board instituted trial, Magna filed a Patent Owner Response
(Paper 26, “PO Resp.”), to which TRW replied (Paper 29, “Pet. Reply”). An
Oral Hearing was held on February 19, 2015, and the Hearing Transcript
(Paper 38, “Tr.”) has been entered in the record.
We have jurisdiction under 35 U.S.C. § 6(c). This Final Decision is
entered pursuant to 35 U.S.C. § 318(a). We determine that TRW has shown
by a preponderance of the evidence that the challenged claims are
unpatentable.
B. Related Proceedings
TRW discloses that the ’462 patent has been asserted in Magna
Electronics, Inc. v. TRW Automotive Holdings Corp., Case No. 1:12-cv-
00654-PLM (W.D. Mich. 2012). Pet. 7; Paper 2, 2.

1 U.S. Patent No. 4,970,653 to Kenue, Ex. 1004.
IPR2014-00266
Patent 7,994,462 B2

3

C. The ’462 Patent (Ex. 1002)
The ’462 patent, titled “Vehicular Image Sensing System,” describes a
system for controlling a vehicle—e.g., dimming the vehicle’s headlights—in
response to detecting “objects of interest” in front of the vehicle—e.g., the
headlights of oncoming vehicles and the taillights of leading vehicles.
Ex. 1002, Abst., 1:22–27. The system uses an image sensor that divides the
scene in front of the vehicle into “a plurality of spatially separated sensing
regions.” Id. at 2:16–19. A control circuit with a processor receives image
data from the image sensor and determines if individual regions include light
sources having particular characteristics, such as a “spectral characteristic”
(color), or intensity. Id. at 1:65–2:9, 3:43–51. By comparing the lights’
characteristics with the distribution of the lights across the regions, such as
the lights’ proximity to each other and to the vehicle’s central axis, the
system can distinguish oncoming headlights and leading taillights from
streetlights and other lights that are not of interest. Id. at 9:32–61, 10:53–56.
The system also may detect traffic signs and lane markers, and assist the
driver in other ways, such as alerting the driver to lane changes. Id. at
11:60–12:13.
D. Illustrative Claims
Claims 30 and 34 are independent, and each is drawn to an image-
sensing system for a vehicle. Ex. 1002, 15:18–47, 15:60–16:22. These
claims share at least three common limitations: (1) an image sensor
comprising a two-dimensional array of light-sensing photosensor elements;
(2) the image sensor being inside the vehicle on which it is mounted, having
a forward field of view through the vehicle’s windshield; and (3) a control
IPR2014-00266
Patent 7,994,462 B2

4

comprising a processor that processes the image data to identify objects of
interest. Id.
Claim 30 is illustrative and is reproduced below:
30. An image sensing system for a vehicle, said image sensing
system comprising:
an imaging sensor comprising a two-dimensional array of
light sensing photosensor elements;
said imaging sensor having a forward field of view to the
exterior of a vehicle equipped with said image sensing system
and through the windshield of the equipped vehicle;
wherein said imaging sensor is operable to capture image
data;
a control comprising an image processor;
wherein said image sensing system identifies objects in said
forward field of view of said image sensor via processing of
said captured image data by said image processor;
wherein said image processing comprises pattern
recognition and wherein said pattern recognition comprises
detection of at least one of (a) a headlight, (b) a taillight and (c)
an object, and wherein said pattern recognition is based at least
in part on at least one of (i) shape, (ii) reflectivity, (iii)
luminance and (iv) spectral characteristic; and
wherein said control at least one of (a) controls a headlamp
of the equipped vehicle as a function of a speed of the equipped
vehicle, (b) controls a headlamp of the equipped vehicle in
response to said image processing, (c) controls a speed of the
equipped vehicle in response to said image processing, and (d)
generates an alert to the driver of the equipped vehicle in
response to said image processing.

II. ANALYSIS
A. Claim Construction
“A claim in an unexpired patent shall be given its broadest reasonable
construction in light of the specification of the patent in which it appears.”
37 C.F.R. § 42.100(b); see In re Cuozzo Speed Tech., LLC, 778 F.3d 1271,
IPR2014-00266
Patent 7,994,462 B2

5

1281 (Fed. Cir. 2015) (“We conclude that Congress implicitly adopted the
broadest reasonable interpretation standard in enacting the AIA.”). Under
that standard, the claim language should be read in light of the specification
as it would be interpreted by one of ordinary skill in the art. In re Suitco
Surface, Inc., 603 F.3d 1255, 1260 (Fed. Cir. 2010). Thus, we generally
give claim terms their ordinary and customary meaning. See In re
Translogic Tech., Inc., 504 F.3d 1249, 1257 (Fed. Cir. 2007) (“The ordinary
and customary meaning is the meaning that the term would have to a person
of ordinary skill in the art in question.”) (internal quotation marks omitted).
We expressly interpret below only those claim terms that require
analysis to resolve arguments related to the patentability of the challenged
claims.
1. “pattern recognition”
In the Decision on Institution we construed “pattern recognition” to
mean “detection of an object of interest based upon shape, reflectivity,
luminance, or spectral characteristic.” Dec. 7. We based this construction
on an express definition of the term in the Specification:
Pattern recognition may be used to further assist in the
detection of headlights, taillights, and other objects of interest.
Pattern recognition identifies objects of interest based upon
their shape, reflectivity, luminance, and spectral
characteristics. For example, the fact that headlights and
taillights usually occur in pairs could be used to assist in
qualifying or disqualifying objects as headlights and taillights.
By looking for a triad pattern, including the center high-
mounted stoplight required on the rear of vehicles, stoplight
recognition can be enhanced.
Ex. 1002, 11:1–10 (emphasis added). We further noted that although the
Specification describes pattern recognition as identifying objects of interest
IPR2014-00266
Patent 7,994,462 B2

6

based on their shape, reflectivity, luminance, and spectral characteristics,
i.e., based on all four of these characteristics, the claims require pattern
recognition to be based on only one of the characteristics. For example,
claim 30 requires the image processing to comprise “pattern recognition . . .
wherein said pattern recognition is based at least in part on at least one of
(i) shape, (ii) reflectivity, (iii) luminance and (iv) spectral characteristic.” Id.
at 15:33–39 (emphasis added). Claim 34 includes similar language. In
addition, the examples of pattern recognition provided in the Specification—
e.g., recognizing headlights and taillights because they occur in pairs—seem
to be limited to the shape of the object of interest rather than to any of its
other characteristics. For these reasons we construed “pattern recognition”
to mean detection of objects of interest based upon their shape, reflectivity,
luminance, or spectral characteristic. Dec. 7.
Neither TRW nor Magna expressly disputes this construction.
Furthermore, we find nothing in the record adduced subsequent to institution
that contradicts it. Therefore, we adopt this construction as our final
construction of this term.
2. “detection” and “identification”
a. The Parties’ Positions
Claims 30 and 34 recite an image sensing system that “identifies
objects” via image “processing,” the image processing comprising “pattern
recognition,” which itself comprises “detection” of, e.g., an object. Magna
argues that “identification” and “detection” have different meanings, and
that the claims require both “functions” to be performed. PO Resp. 7–9.
According to Magna, “detection can be said to mean the discovery of the
IPR2014-00266
Patent 7,994,462 B2

7

presence or existence of something,” whereas “‘[i]dentification’ requires
establishing or indicating or knowing what an object is.” Magna explains:
That identification is not a synonym for detection is evidenced
by the plain and ordinary meaning of the terms. “Detect” is
dictionary-defined to mean “to discover or notice the presence
of” (Merriam-Webster definition of “detect”, Exhibit 2035),
whereas “Identify” is dictionary-defined to mean “to know and
say who someone is or what something is” (Merriam-Webster
definition of “identify”, Exhibit 2036.) Plainly, identification is
different from and more than detection. The presence of an
object may be detected but the object may not necessarily be
identified.
Id. at 7 n.3 (emphasis added). Magna also provides a number of examples of
how these terms are used in the Specification to support its position that the
two terms have different meanings. Id. at 6–9.
TRW disagrees with Magna’s construction. In particular, TRW
disagrees that identification of an object requires something more than
detecting the object. TRW asserts that the claims do not merely recite
detection of an unknown object, but rather recite detection of specific
objects of interest, e.g., a headlight and taillight, in the first instance. Pet.
Reply 3–4 (citing Ex. 1002, claims 30, 34). TRW reasons that
[U]sing Magna’s definitions [for detection and identification],
to ‘detect a headlight’ (i.e., to determine that a headlight is
present) the image sensing system must also ‘identify a
headlight’ (i.e., recognize that the thing detected is a headlight).
(Id.) Identifying an object as a headlight, therefore, does not
“require[] more than” detecting a headlight—identifying a
headlight and detecting a headlight are one and the same.
Id. at 4.
IPR2014-00266
Patent 7,994,462 B2

8

b. Analysis
In determining the ordinary and customary meaning of a claim term,
the Board may consult a general dictionary definition of the word for
guidance. Comaper Corp. v. Antec, Inc., 596 F.3d 1343, 1348 (Fed. Cir.
2010). Based on the dictionary passages submitted by Magna, there is
substantial similarity between the definitions of “detect” and “identify.”
According to these dictionary excerpts, one definition of “detect” is “to
discover the true character of” (Ex. 2035), whereas a definition of “identify”
is “to find out . . . what something is” (Ex. 2036). Given this similarity, it is
understandable that the Specification uses the terms interchangeably. For
example, according to the Specification, the invention “identifies” oncoming
headlights and leading taillights, but it also “detects” oncoming headlights
and leading taillights. Ex. 1002, 3:2–3, 6:3–12, 7:7–8. Likewise, the
Specification describes “pattern recognition” as assisting in the “detection”
of objects of interest, but it also “identifies” objects of interest. Id. at 11:1–
3. Although different terms in a claim are presumed to have different
meanings, that presumption may be rebutted when, as here, the Specification
uses the terms interchangeably. See In re Magna Elecs., Inc., Case Nos.
2014–1798, 2014–1801, 2015 WL 2110525, at *5 (Fed. Cir. May 7, 2015)
(rejecting Magna’s argument that a “positional relationship” has a different
meaning than “an indication of a distance” because the patent “essentially
treats the two terms coextensively”).
Even assuming arguendo that a distinction exists between ”detect”
and “identify,” and that Magna is correct that “the presence of an object may
be detected but the object may not necessarily be identified” (PO Resp. 7
n.3), we disagree with Magna that identification requires more than
IPR2014-00266
Patent 7,994,462 B2

9

detection in the context of the claims at issue. As stated above, claims 30
and 34 recite an image sensing system that “identifies” objects via image
processing, the image processing comprising pattern recognition, the pattern
recognition, in turn, comprising “detection” of, e.g., an object. In other
words, the claimed system identifies objects by detecting them.2 The claim
does not present the terms as two separate requirements (e.g., a system that
“identifies and detects”), but rather is structured such that performing one
satisfies the other. “[I]dentif[ying]” an object is satisfied when the system
“detect[s]” an object, the only additional requirement being that the
detection be the result of “pattern recognition . . . based at least in part on at
least one of (i) shape, (ii) reflectivity, (iii) luminance and (iv) spectral
characteristic.”
B. Kenue
Kenue describes a “computer vision system” that detects lane markers
and obstacles in front of an automobile. Ex. 1004, 1:53–61. Figure 1 of
Kenue, reproduced below, depicts Kenue’s system in block-diagram form:

2 Conversely, claim 1 of U.S. Patent No. 7,459,664 B2 (“the ’664 patent”),
the grand-parent of the ‘462 patent, recites a system that “detects” objects by
“identifying” them—that is, the “detects” and “identifies” terminology is
reversed compared to the analogous language in claims 30 and 34 of the
’462 patent. See IPR2015-00256, Ex. 1002, 13:35–36 (“said sensing system
detecting objects by processing said image data to identify objects”).
Generally, claims in patents in the same family that share a common written
description are interpreted consistently. NTP, Inc. v. Research In Motion,
Ltd., 418 F.3d 1282, 1293 (Fed. Cir. 2005), abrogated on other grounds by
Zoltek Corp. v. United States, 672 F.3d 1309, 1323 (Fed. Cir. 2012) (en
banc). The reversal of terms between claim 1 of the ’664 patent and claims
30 and 34 of the ‘462 patent lends support to the notion that the drafter of
these claims considered the terms to be interchangeable.
IPR2014-00266
Patent 7,994,462 B2

10

As depicted in Figure 1, Kenue’s system comprises black and white CCD
[charge coupled device] video camera 10—mounted on a vehicle’s
windshield to capture the driver’s view of the road in front of the vehicle—
coupled to computer 14 via analog-to-digital converter 12. Id. at 2:28–34,
Fig. 1. Computer 14 processes the image data received from the CCD
camera, and drives output devices—i.e., display 16, obstacle warning alarm
18, and utilization circuit 20—in response to the data. Id. at 2:34–48, Fig. 1.
The “raw image” from the camera is “digitized . . . into a 512x512x8
image.” Id. at 3:44–45. Computer 14 receives the digitized image data from
the camera and, using one of two main algorithms, template matching and a
Hough transform, “dynamically define[s] the search area for lane markers
based on the lane boundaries of the previous [image] frame, and provide[s]
estimates of the position of missing markers on the basis of current frame
and previous frame information.” Id. at 2:32–48. The system also detects
and alerts the driver to obstacles in the lane within about 50 feet of the
vehicle. Id. at 2:48–51.
C. Claim 30—Anticipation—Kenue
TRW asserts that Kenue anticipates claim 30. Its contentions in this
regard are summarized in the claim chart at pages 14–17 of the Petition.
Magna disputes that claim 30 is anticipated. In particular, Magna argues
IPR2014-00266
Patent 7,994,462 B2

11

that Kenue fails to disclose (1) a “two-dimensional array of light sensing
photosensor elements,” (2) both “detection” and “identification,” and (3)
“pattern recognition” based on either “shape” or “luminance.” PO Resp. 10–
19. Magna further asserts that Kenue does not identify objects in “the
forward field of view of [the] sensor,” and does not “process[] . . . image
data.” Id. at 20–23.
1. “an imaging sensor comprising a two-dimensional array
of light sensing photosensor elements”
Claim 30 recites “an imaging sensor comprising a two-dimensional
array of light sensing photosensor elements.” TRW identifies Kenue’s CCD
video camera as corresponding to the claimed imaging sensor. Pet. 14. In
describing a block diagram of its system depicted in Figure 1, Kenue states
that it uses a “black and white CCD video camera”; the raw image from the
camera is “digitized . . . into a 512x512x8 image.” Ex. 1004, 2:27–30, 3:44–
46, Fig. 1. This refers to a two-dimensional image that is 512 pixels high
and 512 pixels wide, with each pixel represented by eight bits. Ex. 1020 ¶ 4.
TRW argues that this shows that Kenue uses a two-dimensional array of
light sensing photosensor elements to capture the raw image. Pet. 14.
Magna disagrees. First, Magna argues that “a CCD camera is not
explicitly, necessarily or inherently two-dimensional.” PO Resp. 11 (citing
Ex. 2032 ¶¶ 35, 36–46). Magna notes, for example, that “a line-scan camera
and a printer scanhead use a one-dimensional CCD.” Id. (citing Ex. 2032
¶ 46). Second, Magna argues that “Kenue never discloses explicitly,
necessarily or inherently that forming a two-dimensional image requires use
of a two-dimensional sensor.” Id. Rather, Magna argues, “an image may be
represented by a multidimensional matrix without the image sensor itself
IPR2014-00266
Patent 7,994,462 B2

12

utilizing a two-dimensional array of light sensing photosensor elements.”
Id. According to Magna, “Kenue’s search functions and algorithms could
also be just as easily implemented using data from a one-dimensional array
of light sensors.” Id. at 12–13 (citing Ex. 2032 ¶¶ 51–53).
TRW responds that a person of ordinary skill would understand from
Kenue’s discussion of the creation of a two-dimensional digital image that
Kenue uses a two-dimensional image sensor to capture the raw image from
which the digital image is created. Citing its declarant, Jeffrey A. Miller,
Ph.D., TRW asserts that “[b]y pointing out that the digitized image is 512 x
512 x 8, Kenue explicitly teaches to the skilled artisan use of a two-
dimensional imaging array.” Pet. Reply 2 (citing Ex. 1020 ¶ 6). According
to TRW, it would be difficult, if not impossible, to create a two-dimensional
512x512 image in real time using, e.g., a 1x512 sensor array, because such a
sensor would have to take “512 distinct images . . . [that] would need to be
individually pieced together.” Id. at 1–2. TRW further notes that Kenue
describes digitizing a single “raw image” into a 512x512 image, rather than
multiple “raw images,” indicating that the patent does not contemplate such
piecing together of multiple raw images as would be required if a one-
dimensional sensor were used. Id. at 2. Furthermore, TRW contends that
“Magna’s own expert . . . admits Kenue does indeed teach a two
dimensional image sensor.” Id.
“A claim is anticipated only if each and every element as set forth in
the claim is found, either expressly or inherently described, in a single prior
art reference.” Verdegaal Bros. v. Union Oil Co. of California, 814 F.2d
628, 631 (Fed. Cir. 1987). However, “the reference need not satisfy an
ipsissimis verbis test.” In re Gleave, 560 F.3d 1331, 1334 (Fed. Cir. 2009)
IPR2014-00266
Patent 7,994,462 B2

13

(internal quotation marks omitted). Rather, the reference must provide
sufficient description of the claim elements so that a person of ordinary skill
in the art would understand them to be present. See Akzo N.V. v. U.S. Int’l
Trade Comm’n, 808 F.2d 1471, 1479 (Fed. Cir. 1986) (stating that the
Commission did not apply an improper ipsissimis verbis test in its
anticipation analysis, but rather considered whether the prior art reference
disclosed the claimed invention to one of ordinary skill in the art); In re
Paulsen, 30 F.3d 1475, 1480 (Fed. Cir. 1994) (holding that prior art
references must be “considered together with the knowledge of one of
ordinary skill in the pertinent art”). Thus, it is not dispositive that Kenue
does not state that it uses a “two-dimensional array of light sensing
photosensor elements” in so many words, as long as a person of ordinary
skill would understand from the disclosure that the reference is describing
such structure.
We find that Kenue discloses to a person of ordinary skill that its
black and white CCD video camera system uses a two-dimensional array of
photosensor elements. First, we credit the testimony of Dr. Miller that a
person of ordinary skill would understand that, because Kenue’s system
creates a two-dimensional digital image from a raw image, the system uses a
two-dimensional photosensor array to capture the raw image. See Ex. 1020
¶ 6. Further, we rely on Kenue’s disclosure that it digitizes a single raw
image, rather than multiple raw images, to create the 512x512x8 digital
image, indicating to a person of ordinary skill that the single raw image
would had to have been captured by a 512x512 array of photosensor
elements. Finally, although Magna’s declarant, Matthew A. Turk, Ph.D.,
originally testified that Kenue’s CCD camera does not “necessarily” include
IPR2014-00266
Patent 7,994,462 B2

14

a two-dimensional array of photosensor elements (Ex. 2032 ¶ 36), and that
“Kenue never discloses explicitly, necessarily or inherently that forming a
two-dimensional image requires use of a two-dimensional sensor” (id. ¶ 47),
Dr. Turk provided testimony in his deposition that supports the opposite
conclusion. Dr. Turk noted that the “camera image plane” depicted in
Figure 2 “correspond[s] to the image sensor in Kenue,” and that a plane is “a
two-dimensional surface.” Ex. 1013, 198:1–13.
2. “said imaging sensor having a forward field of view to
the exterior of a vehicle equipped with said imaging
sensing system and through the windshield of the
equipped vehicle”
Kenue’s CCD video camera is “mounted in a vehicle, say at the upper
center of the windshield to capture the driver’s view of the road ahead.”
Ex. 1004, 2:29–32. We agree with TRW that this teaching corresponds to
the claimed requirement that the imaging sensor has “a forward field of view
to the exterior of a vehicle . . . and through the windshield of the equipped
vehicle.”
3. “said imaging sensor is operable to capture image data”
Kenue teaches that its computer “is programmed with algorithms for
processing the images sensed by the camera.” Id. at 2:40–41. We agree
with TRW that this teaching corresponds to the claimed requirement that the
imaging sensor is operable to capture image data.
4. “a control comprising an image processor”
As depicted in Figure 1, Kenue’s system comprises computer 14, and
“output devices driven by the computer.” Id. at 2:33–35. We agree with
TRW that this teaching corresponds to the claimed requirement for a control
that comprises an image processor.
IPR2014-00266
Patent 7,994,462 B2

15

5. “wherein said image sensing system identifies objects in
said forward field of view of said image sensor via
processing of said captured image data by said image
processor”
TRW argues that Kenue teaches this limitation because it identifies
lane markers and obstacles in the forward field of view of the equipped
vehicle by digitizing raw image data and processing the digital image using
either a template-matching algorithm or a Hough transform. Pet. 15 (citing
Ex. 1004, 2:41–51, 3:44–46, 59–66).
Magna makes two arguments with respect to this limitation. First,
Magna argues that Kenue does not identify objects in the “forward view of
[the] sensor.” According to Magna, Kenue is limited to searching for lane
markers only in a restricted view, and “‘forward field of view’ means more
than th[is] restricted view.” PO Resp. 20–21. Magna derives this position
from the fact that the Specification of the ’462 patent teaches that the
invention is capable of “evaluation of light source characteristics made in
each portion of the scene forward of the vehicle.” Id. (citing Ex. 1002,
1:65–67, 2:1–4).
As an initial matter, it is unclear from Magna’s Response whether it
believes that this claim limitation requires that all objects present in the
camera’s field of view be identified, or whether it just requires something
more than searching for lane markers in Kenue’s limited search area. In any
event, we do not find support in either the claim language or the
Specification for Magna’s position. The limitation simply requires that
“objects” be identified in the forward field of view of the image sensor, but
does not otherwise specify what objects must be identified or where in the
field of view the objects must be found (there does not seem to be any
IPR2014-00266
Patent 7,994,462 B2

16

dispute that a lane marker is an “object” or that Kenue’s limited search areas
are in the forward view of view of Kenue’s system). As the limitation does
not expressly require more than this, we decline Magna’s invitation to read
additional requirements into the claim.
Second, Magna argues that “Kenue does not ‘process[] . . . data’
because it does no more than convert analog to digital.” PO Resp. 22. TRW
responds that Kenue’s computer 14 processes data as required by the claim.
Pet. Reply 13. We agree. Kenue’s computer 14 receives digitized image
data and processes it to detect lane markers using one of two algorithms:
template matching or a Hough transform. Ex. 1004, 2:40–44. This
corresponds to the claimed processing of image data.
6. “wherein said image processing comprises pattern
recognition, and wherein said pattern recognition
comprises detection of at least one of (a) headlight, (b) a
taillight and (c) an object, and wherein said pattern
recognition is based at least in part on at least one of (i)
shape, (ii) reflectivity, (iii) luminance and (iv) spectral
characteristics”
TRW asserts that Kenue discloses “a template matching algorithm and
a Hough transform algorithm for detecting lane markers (i.e., ‘an object’)[,
and b]oth these image processing algorithms comprise pattern recognition
and are based at least in part on at least one of shape and luminance.” Pet.
15–16.
Magna disputes that Kenue discloses this limitation, making two
arguments in this regard. First, Magna argues that Kenue does not disclose
both the “detection” and “identification” functions. According to Magna,
“TRW collapses [‘detection’ and ‘identification’] into one [requirement] and
does not explain how Kenue’s teachings ‘identify’ objects.” PO Resp. 13.
IPR2014-00266
Patent 7,994,462 B2

17

According to Magna, “[t]his is because Kenue does not need to nor does it
identify objects as all of Kenue’s teachings are only specific to lane
markers.” Id.
TRW responds that the “skilled artisan would understand that
detection and identification do not require the imaging system to perform
two distinct functions.” Pet. Reply 6. TRW further argues that Kenue’s
system identifies lane markers (i.e., more than one lane marker), and also
identifies “objects as obstacles.” Id. According to TRW, the “claims do not
require that the identified objects be identified with greater precision.” Id.
Magna’s argument is based on its contention that this limitation
requires the system to both “identify” and “detect” objects, i.e., that
identification and detection are two separate required “functions.” As
discussed above, however, we disagree with this construction. Instead, this
limitation makes clear that, as recited in claim 30, detection of an object
satisfies the requirement that an object be identified. Here, Kenue clearly
teaches detection of lane markers and obstacles, both of which are
indisputably objects. See, e.g., Ex. 1004, 1:59–62 (“The invention is . . . a
method of detecting lane markers.”); id. at 2:49–51 (“preprocessing
procedures detect obstacles in the lane within about 50 feet of the vehicle”);
id. at 3:3–4 (“broken line boxes 28 around the markers 24 define the area to
be searched for detecting markers”).
Magna argues that “[a] system that only cares about one object does
not need to, nor would it, identify objects as is recited in the claims.” PO
Resp. 15. This argument implicitly reads into the claim the requirement that
the system be able to identify more than one “type” of object; e.g., a lane
marker and something other than a lane marker. But the claim language
IPR2014-00266
Patent 7,994,462 B2

18

contains no such requirement. It merely requires that the system identifies
“objects,” which Kenue satisfies by identifying lane markers. Even if this
limitation did contain such a requirement, Kenue would satisfy it because it
detects both lane markers and “obstacles,” e.g., other vehicles, both of which
would correspond to the claimed “object.”
Second, Magna argues that neither Kenue’s template matching
algorithm nor its Hough transform discloses pattern recognition based on
either shape or luminance. PO Resp. 16–20. Magna argues that the term
“shape” in this limitation only “refers to the object of interest’s shape.” Id.
at 18. According to Magna, “Kenue only checks edges of a pre-processed,
digitized, and altered image,” or “the shape of a Hough transform,” but
“does not detect the shape of an object or pattern.” Id. (citing Ex. 1004,
6:30–46; Ex. 2032 ¶ 73). Magna further argues that “intensity,” which is
“the measurable amount of a property,” is not the same as “luminance,”
which “describes the amount of light that passes through or is emitted from a
particular area.” Id. at 19 (citing Ex. 1004, 4:49–66; Ex. 2032 ¶ 74).
According to Magna, TRW has not explained how the correlation of a
window of constant “gray level” with the “gray level” of the image relates to
luminance. Id.
TRW responds that Dr. Turk admitted in his deposition that template
matching is a form of pattern recognition, and that Kenue teaches that in
template matching “‘a template or window of desired intensity and shape is
correlated with the image to create a correlation matrix’ which is
subsequently used to identify lane markers.” Pet. Reply 9–10 (citing Ex.
1004, 3:23–26). In addition, TRW asserts that “shape” may refer to more
than the shape of the object of interest, and that Kenue’s teaching of using
IPR2014-00266
Patent 7,994,462 B2

19

the desired shape of its template window falls within the scope of this
limitation. Id. at 12 (citing Ex. 1002, claims 30 and 34; Ex. 1004, 3:23–26;
Ex. 1020 ¶ 19). TRW further argues that the “lines (and the points to which
they are transformed in the Hough space) have shapes, and these shapes are
specifically utilized in the calculus.” Id.
We have reviewed the record, including the declarations submitted by
Dr. Turk and Mr. Miller, and determine that TRW has established by a
preponderance of the evidence that Kenue teaches pattern recognition based
at least in part on shape. First, as Dr. Turk acknowledges, template
matching is a type of pattern recognition. Ex. 1013, 51:4–12. Second, we
agree with TRW that “shape” is not limited to the shape of the object being
detected; rather, this limitation simply requires that pattern recognition be
based at least in part on shape. Ex. 1002, claim 30. Therefore, “shape”
encompasses the shape of the template window that is used to detect the
object when the template-matching algorithm is used. Ex. 1020 ¶ 19.
Finally, we note that lane markers are detected at least in part based on their
shape (and not only based on the shape of the template window). For
example, the fact that the shape of the lane markers causes them to appear
comprised of diagonal lines is used to distinguish the lane markers from
obstacles, which are assumed to be comprised primarily of vertical and
horizontal lines. Ex. 1004, 4:11–17. The Hough transform algorithm also is
based at least in part on the shape of the lane markers, because as part of that
process the edges and contour of the lane markers are determined. Id. at
5:65–6:18, Fig. 6a.
IPR2014-00266
Patent 7,994,462 B2

20

7. “wherein said control at least one of (a) controls a
headlamp of the equipped vehicle as a function of a
speed of the equipped vehicle, (b) controls a headlamp of
the equipped vehicle in response to said image
processing, (c) controls a speed of the equipped vehicle
in response to said image processing, and (d) generates
an alert to the driver of the equipped vehicle in response
to said image processing.
TRW argues that Kenue describes option (d) of this limitation, i.e., it
describes that the control generates an alert to the driver of the equipped
vehicle in response to the image processing. Pet. 17. TRW refers to the fact
that Kenue’s system alerts the driver when an obstacle is detected in the road
less than 50 feet away from the equipped vehicle. Id. (citing Ex. 1004,
Abst., 2:34–39). Magna does not dispute this contention specifically. We
agree with TRW and find that Kenue’s teaching of alerting the driver upon
the detection of an obstacle less than 50 feet from the vehicle corresponds to
option (d) of the above limitation. Accordingly, we find that Kenue
anticipates claim 30.
D. Claim 31
Claim 31 depends from claim 30 and additionally recites “wherein
said imaging sensor is operable to capture frames of image data and wherein
identification of objects is enhanced by comparing image data of objects
over successive frames of said captured image data.” TRW asserts that
Kenue describes these additional limitations, and relies on Kenue’s
teachings that (1) Kenue’s algorithms “dynamically define the search area
for lane markers based on the lane boundaries of the previous frame, and
provide estimates of the position of missing markers on the basis of current
frame and previous frame information” (Ex. 1004, 2:44–48); (2) Kenue’s
IPR2014-00266
Patent 7,994,462 B2

21

“search area changes with marker position MP, and increases in size if the
marker was not found in the search of the previous frame” (id. at 4:27–29);
and (3) “[a]fter the centroid of each marker is calculated . . ., if the markers
are not found where expected . . . based on the previously detected lane
geometry, the marker locations from the previous frame are used” (id. at
5:14–18). Pet. 17–18.
Magna responds that Kenue does not disclose that “identification of
objects is enhanced” by comparing image data of objects over successive
frames, as recited in dependent claim 31. According to Magna, “Kenue does
not enhance object recognition by comparing identified objects over
successive frames[, but rather] detects lane markers and compares the
position of the lane markers to estimate lane boundaries.” PO Resp. 27.
We agree with TRW, however, that Kenue describes this limitation.
Kenue’s teaching that it compares a “previous frame” with a “current frame”
corresponds to “comparing image data of objects over successive frames of
said captured image data.” Further, the comparison of previous-frame and
current-frame image data “dynamically define[s]” the search area for lane
markers. That is, the area where subsequent lane markers are expected to be
found is determined by where previous lane markers were found. This
increases the chances of making valid identifications of lane markers by
guiding the search to areas where lane markers are likely to appear, and
therefore “enhances” the identification of lane markers. Accordingly, we
find that Kenue anticipates claim 31.
E. Claim 34
Claim 34 is largely identical to Claim 30 except that it replaces the
last limitation of claim 30 with the following:
IPR2014-00266
Patent 7,994,462 B2

22

wherein said image sensing system is operable to at least one of
(a) detect lane markers on a road being traveled by the equipped
vehicle and present in the field of view of said imaging sensor
in order to at least one of (i) assist the driver in steering the
equipped vehicle and (ii) provide a warning to the driver of the
equipped vehicle, and (b) identify traffic signs.
TRW argues that Kenue’s system “is operable to detect the lane markers on
the road being traveled by the equipped vehicle and present in the field of
view of the equipped vehicle to both, provide a warning to the driver of the
equipped vehicle, and to assist the driver in steering the equipped vehicle.”
Pet. 32–33 (citing Ex. 1004, 2:27–39, 5:5–6, 7:65–66). Magna does not
dispute this contention. Having reviewed the cited portions of Kenue, we
agree with TRW that they describe this limitation. Having previously found,
in the context of discussing claim 30, that Kenue describes the remaining
limitations of claim 34, we find that Kenue anticipates claim 34.
F. Claim 37
Claim 37 depends from claim 34 and additionally recites “wherein
said imaging sensor is operable to capture frames of image data and wherein
identification of objects is enhanced by comparing image data of objects
over successive frames of said captured image data.” We previously found
that Kenue describes this limitation in the context of discussing claim 31,
and that Kenue anticipates claim 34. Accordingly, we find that Kenue
anticipates claim 37.
G. Claim 38
Claim 38 depends from claim 34 and additionally requires the control
to, e.g., “generate[] an alert to the driver of the equipped vehicle in response
to said image processing.” We found that Kenue describes this limitation in
IPR2014-00266
Patent 7,994,462 B2

23

the context of discussing claim 30. Therefore, having previously found that
Kenue anticipates claim 34, we find that Kenue anticipates claim 38.
III. CONCLUSION
For the foregoing reasons, we determine that TRW has shown, by a
preponderance of the evidence, that Kenue anticipates claims 30, 31, 34, 37,
and 38.
IV. ORDER
For the reasons given, it is
ORDERED that claims 30, 31, 34, 37, and 38 of the ’462 patent are
determined to be unpatentable.

This is a Final Decision. Parties to the proceeding seeking judicial
review of the decision must comply with the notice and service requirements
of 37 C.F.R. § 90.2.

IPR2014-00266
Patent 7,994,462 B2

24

PETITIONER:

Josh C. Snider
Timothy K. Sendek
A. Justin Poplin
LATHROP & GAGE LLP
patent@lathropgage.com
jsnider@lathroppage.com
tsendek@lathropgage.com
jpoplin@lathropgage.com

PATENT OWNER:

Timothy A. Flory
Terence J. Linn
GARDNER, LINN, BURKHART & FLORY, LLP
Flory@glbf.com
linn@glbf.com

David K.S. Cornwell
Jason Eisenberg
Robert Sterne
STERNE, KESSLER, GOLDSTEIN & FOX PLLC
Davidc-PTAB@skgf.com
jasone-PTAB@skgf.com
rsterne-PTAB@skgf.com