`____________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`____________________
`
`SAMSUNG ELECTRONICS CORPORATION, LTD., AND
`SAMSUNG ELECTRONICS AMERICA, INC.,
`Petitioners
`
`v.
`
`IMAGE PROCESSING TECHNOLOGIES, LLC,
`Patent Owner
`____________________
`
`CASE IPR2017-01218
`Patent No . 8,983,134 B2
`____________________
`
`
`
`
`
`PATENT OWNER IMAGE PROCESSING TECHNOLOGIES LLC’S
`RESPONSE TO PETITIONER’S SUPPLEMENTAL BRIEF
`
`Paper No. 34
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`
`I.
`
`The Natural and Correct Reading of Element [1c] Requires that
`Determining Target Boundaries Must Occur as a Part of the
`Formation of the Histogram. .......................................................................... 1
`The Specification Supports Image Processing’s Construction ...................... 2
`II.
`III. The Prosecution History Supports Image Processing’s Construction ............ 5
`A.
`Samsung’s Reading of Element [1c] and Claim 4 Encompasses
`Hunke and Grove, Over Which the Claims Were Allowed ................. 5
`The Examiner Distinguished between Processing in a Frame
`and Processing Over Multiple Frames ................................................. 7
`
`B.
`
`
`
`
`
`
`
`
`
`
`
`The Board authorized Image Processing Technologies, LLC (“Image
`
`Processing”) to file this response. Paper 31 (Order), 2 (stating that Petitioner’s
`
`brief raised new claim interpretation issues). The correct claim construction of
`
`element [1c] under Phillips is dispositive of this IPR because the Petition and its
`
`exhibits do not show that the prior art teaches or suggests this claim element, (see
`
`Paper 11 (Institution Decision), 18 (Gerhardt), 28 (Gilbert, Hashima)), and no
`
`additional evidence or argument has been authorized (Paper 26 (Order), 4).
`
`The IPR2017-00353 panel did not have the benefit of a full record, e.g., Ex.
`
`2011 (Hart deposition), so that Panel’s claim interpretation should not control.
`
`I. The Natural and Correct Reading of Element [1c] Requires that
`Determining Target Boundaries Must Occur as a Part of the Formation
`of the Histogram.
`Under Phillips, the Board attempts to identify the correct construction in
`
`light of the claim language, specification, and prosecution history, not the broadest
`
`construction. PPC Broadband v. Corning Optical Commc’ns, 815 F.3d 734, 740
`
`(Fed. Cir. 2016) (under Phillips, the PTO should seek “the construction that most
`
`accurately delineates the scope of the claim invention”). Under Phillips, the
`
`meaning most closely aligned with the plain language of claim element [1c]
`
`“forming the…histogram…comprises determining…boundaries of the target,”
`
`requires that determining the boundaries be part of forming the histogram.
`
`Samsung’s position (Paper 29, 2–3, 6) reads element [1c] unreasonably
`
`1
`
`
`
`broadly as encompassing an unlimited amount of post-histogram-formation
`
`activity prior to determination of the target boundaries.1 See Ex. 2011, 114:14–
`
`115:23. As Dr. Hart stated in deposition:
`
`Q Is there any limitation on how much additional processing can be
`done after the histogram is formed in order to find boundaries?
`
`A I don't see any limitation on the amount of computation or analysis.
`I think ’134, Claim 1 and specifically Element 1C says that you form
`a histogram and determine the X and Y minima and maxima as
`boundaries of the target. And I think that [if] determination is based
`on the formation of that histogram . . . then you satisfied the
`restrictions of Element 1C.
`
`Ex. 2011, 115:11–23. Allowing an unlimited amount of post histogram-formation
`
`processing reads out the “comprising” language, and instead merely requires both a
`
`“forming” step and a “determining” step with no relationship between the steps.
`
`II. The Specification Supports Image Processing’s Construction
`Contrary to Samsung’s assertion, the lock-on tracking embodiment (Ex.
`
`1001, 23:59–25:2, Figs. 20–23) is an embodiment of claims 1–6 that teaches
`
`iteratively adjusting a selected area while forming a histogram such that the X and
`
`Y minima and maxima of boundaries of a target are determined as part of forming
`
`
`1 Phillips requires consideration of claim language, specification, and prosecution
`history. The Phillips doctrine of construing claims to preserve their validity
`applies to AIA trials. See Changes to the Claim Construction Standard for
`Interpreting Claims Before the PTAB, 83 FED. REG. 21221, 21223 (May 9, 2018).
`2
`
`
`
`the histogram. See Paper 15, 8–13. The lock-on tracking embodiment is
`
`consistent with dependent claims 4–6. The embodiment teaches setting boundaries
`
`in the x and y histogram formation units 28 and 29 such that only the pixels falling
`
`in the bounded area will be processed. Ex. 1001, 24:1–12, 35–54. The
`
`embodiment teaches processing “successively larger areas” and “adjusting the
`
`center of the area” for which pixels are processed based upon the shape of the
`
`object. Ex. 1001, 24:1–12. Pixel data from an expanded area area can be added to
`
`a histogram of the smaller original area. See Ex. 2011 (Hart depo.), 79:14–83:24.
`
`Samsung’s interpretation of claim 6 is incorrect. Paper 29, 2. Claim 6 is
`
`consistent with the lock-on tracking embodiment and merely requires setting X and
`
`Y value boundaries in the histogram formation units to adjust the selected area.
`
`This is how the lock-on tracking embodiment changes the area in the box shown in
`
`Figures 21–22 while the histogram is being calculated. Ex. 1001, 24:1–12, 35–54.
`
`Thus, Samsung’s argument that the lock-on tracking embodiment must
`
`operate across multiple frames is meritless. Samsung claims that an iterative
`
`process must form a new histogram for each iteration (Paper 29, 4) but Dr. Hart
`
`admitted that a single histogram can continue to be formed if the area under
`
`consideration is enlarged to include more pixel data. Ex. 2011, 79:14–83:24; Ex.
`
`2014. Samsung’s argument that the histogram memory is “cleared” only between
`
`frames thus misses the point. The histogram memory need not be cleared for
`
`3
`
`
`
`multiple iterations within a single frame. Samsung’s argument that the tracking
`
`box is updated frame-by-frame (Paper 29, 4) also misses the point because an
`
`iterative process may execute for each frame, resulting in frame-by-frame updates.
`
`Samsung’s assertion that a claim that limits adjustments to within a single
`
`frame “contradicts” the patent’s teachings (Paper 29, 5) is incorrect. The
`
`specification distinguishes between (i) a video conference embodiment (Figures
`
`15–19, 22:4–23:58) for tracking a person in a video conference, including tracking
`
`the user’s head, Ex. 1001, 22:19–23:34 (see Paper 15, 35); and (ii) the lock-on
`
`tracking embodiment (Figures 20–23, 23:59–25:2) for tracking a simulated object
`
`(Figures 20–21, item 218) selected by the user via the mouse. (See Paper 15, 37.)
`
`The lock-on tracking embodiment does not incorporate post histogram-formation
`
`analysis such as Figure 17, and uses DP=1 as an “example.” Ex. 1001, 24:13–15.
`
`Samsung’s interpretation of the specification (Paper 29, 5) also ignores and
`
`contradicts the prosecution history. During prosecution of the parent ’001 patent
`
`(as discussed in Section III.B, below) separate claims were proposed that were
`
`directed to each of the two embodiments described above. The videoconference
`
`embodiment, e.g. Figure 17, describes forming histograms and then analyzing the
`
`histogram peaks, in order to determine the center of face V “at regular intervals,”
`
`(Ex. 1001, 23:23–24), which is like claim 59 that was proposed (and rejected)
`
`claim 59 during prosecution of the ’001 patent (see page 8, below). The lock-on
`
`4
`
`
`
`tracking embodiment is an iterative process of finding boundaries during histogram
`
`formation for a single frame, which is like claims 58 and 63 that were proposed
`
`(and deemed patentable) during prosecution of the ’001 patent.
`
`III. The Prosecution History Supports Image Processing’s Construction
`The prosecution histories of the ’134 and ’001 patents (same specification
`
`and Examiner) show that even under the BRI standard the Examiner did not read
`
`element [1c] as broadly as Samsung asserts (Paper 29, 2–3, 6). Also, contrary to
`
`Samsung’s argument (id., 5), the content of allowed and rejected claims shows that
`
`the Examiner distinguished between processing during histogram formation within
`
`a frame and post histogram-formation processing across multiple frames.
`
`A. Samsung’s Reading of Element [1c] and Claim 4 Encompasses
`Hunke and Grove, Over Which the Claims Were Allowed
`The Examiner indicated the element [1c] is allowable over Hunke during
`
`prosecution of the ’134 patent. Paper 15, 18–19. During prosecution of the ’001
`
`patent, the Examiner rejected then-pending independent claim 51 but indicated that
`
`dependent claims 55 (containing the language of ’134 patent claim 4) and 56
`
`(containing the language of ’134 patent element [1c]) were allowable over Hunke:
`
`51. [Identical language as ’134 Patent claim 1 elements 1[pre], 1[a]]…
`identifying the target from said at least one histogram.
`
`52. The process according to claim 51 further comprising drawing a
`tracking box around the target.
`
`55. The process according to claim 51, wherein forming the at least
`5
`
`
`
`one histogram further comprises successively increasing the size of a
`selected area until the boundary of the target is found.
`
`56. The process according to claim 52, wherein forming the at least
`one histogram further comprises determining X minima and maxima
`and Y minima and maxima of boundaries of the target.
`
`Ex. 1022, 140–47 (June 23, 2011 Final Office Action), 153–61 (Amendment).
`
`Yet, under Samsung’s interpretation, claim elements [1c] and claim 4
`
`encompass Hunke. Hunke identifies a target using a histogram, then finds target
`
`boundaries post-histogram, like Samsung’s asserted art (Paper 29, 6–8). The
`
`tracking box for Hashima is based on post histogram-formation analysis using
`
`target center and total X, Y length. Ex. 1006, 14:4–12. The Hashima tracking box
`
`is adjusted for the next, not the same, frame (Paper 8 at 33–35; Ex. 1006, 5:66–67,
`
`Fig. 27), contrary to Samsung’s implication (Paper 29, 7). Gerhardt’s pixel
`
`selection and region-growing also happens post-histogram (Paper 15, 40, 56).
`
`Hunke identifies a target by first finding target colors by analyzing a color
`
`histogram to determine what colors occur more frequently in the target. Ex. 2005,
`
`8:50–9:15, 9:25–30, 9:55–10:2. The color analysis happens after the histogram is
`
`formed. Id., 9:25–30, 10:29–31. Hunke uses color and motion analysis, and also
`
`complex analysis such as pattern classification to “unequivocally” detect the target.
`
`Id., 11:29–47, 13:41–59, 14:38–42. Hunke finds the specific coherent (contiguous)
`
`set of pixels that belongs to the target. Id., 13:28–40.
`
`6
`
`
`
`Hunke’s “unequivocal” detection of target pixels clearly includes defining
`
`the X and Y minima and maxima of the target. Also, Hunke’s target tracking
`
`would require the target dimensions. The tracking module sets a virtual camera
`
`box to leave “enough of a border that the object will remain within the virtual
`
`camera in the next video frame.” Id., 7:1–5. The virtual camera is “constantly
`
`adjusted to the position and size of the tracked target.” Id., 12:29–30.
`
`The Examiner would have at least combined Hunke with Grove to issue an
`
`obviousness rejection if Samsung were correct. The Examiner identified page 3 of
`
`Grove as teaching identifying the target in the histogram itself. Ex. 1022, 65–66.
`
`Page 3 also shows tracking boxes centered on and drawn around the target. Ex.
`
`2006, 3. So the Examiner must have understood element [1c] to mean more than
`
`separately forming, identifying, and subsequently determining boundaries.
`
`B. The Examiner Distinguished between Processing in a Frame and
`Processing Over Multiple Frames
`The pattern of rejections and allowances during ’001 patent prosecution
`
`shows that the Examiner (who also examined the ’134 patent) distinguished
`
`between processing during histogram formation that takes place within a frame
`
`(like the lock-on tracking embodiment), and post-histogram processing that takes
`
`place across multiple frames (like the video conference embodiment).
`
`During ’001 patent, applicant cancelled claims 51–56 (shown above) and
`
`presented an independent claim 57 with language identical to ’134 patent claim 1
`7
`
`
`
`except excluding element 1[c], with dependent claims including 58–59 and 63:
`
`58. The process according to claim 57, wherein identifying the target
`in said at least one histogram further comprises determining a center
`of the target to be between X and Y minima and maxima of the target.
`
`59. The process according to claim 57, wherein identifying the target
`in said at least one histogram further comprises determining the center
`of the target at regular intervals.
`
`63. The process according to claim 57, wherein forming the at least
`one histogram further comprises successively increasing the size of a
`selected area until the boundary of the target is found.
`
`Ex. 1022, 118–26 (August 23, 2011 Amendment) (rejected claim in red).
`
`The Examiner rejected claims 57 and 59 as obvious over the combination of
`
`Hunke and Grove but indicated that dependent claims 58 and 63 would be
`
`allowable. Id., 62–70 (August 15, 2013 Office Action) at 65–66, 69; see Ex. 2006,
`
`3. The Examiner would have also rejected claims 58 and 63 if claim element [1c]
`
`and claim 4 encompassed post histogram-formation processing or processing over
`
`multiple frames as Samsung suggests. Rejection of claim 59 (determining center at
`
`regular intervals) but not claim 58 (identifying comprises finding center between
`
`X-Y target boundaries) shows that Samsung’s interpretation of the specification nis
`
`wrong (Paper 29, 3–5), as the Examiner distinguished between claims for
`
`processing in a single frame and processing for across multiple frames.
`
`Thus, the claim, specification, and prosecution support Patent Owner.
`
`8
`
`
`
`
`
`Dated: June 18, 2018
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`/s/ Chris J. Coulson
`Chris J. Coulson (Reg. No. 61,771)
`BUNSOW DE MORY LLP
`101 Brambach Rd.
`Scarsdale, NY 10583
`Tel.: (646) 502-6973
`ccoulson@bdiplaw.com
`
`Michael Zachary (pro hac vice)
`mzachary@bdiplaw.com
`Lauren N. Robinson (Reg. No. 74,404)
`lrobinson@bdiplaw.com
`Craig Y. Allison (Reg. No. 38,067)
`callison@bdiplaw.com
`
`
`
`
`
`CERTIFICATE OF SERVICE
`
`Pursuant to 37 C.F.R. § 42.6(e), the undersigned certifies that on June 18, 2018, the
`foregoing PATENT OWNER IMAGE PROCESSING TECHNOLOGIES
`LLC’S RESPONSE TO PETITIONER’S SUPPLEMENTAL BRIEF was
`served via electronic mail upon the following counsel of record for the Petitioner:
`
`
`John Kappos (Reg. No. 37,861)
`jkappos@omm.com
`
`Marc J. Pensabene (Reg. No. 37,416)
`mpensabene@omm.com
`
`Nicholas J. Whilt (Reg. No. 72,081)
`nwhilt@omm.com
`
`Brian M. Cook (Reg. No. 59,356)
`bcook@omm.com
`
`Clarence Rowland (Reg. No. 73,775)
`crowland@omm.com
`
`IPTSAMSUNGOMM@OMM.COM
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`/s/ Chris J. Coulson
`Chris J. Coulson (Reg. No. 61,771)
`BUNSOW DE MORY LLP
`101 Brambach Rd.
`Scarsdale, NY 10583
`Tel.: (646) 502-6973
`ccoulson@bdiplaw.com
`
`
`
`