`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`____________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`____________________
`
`SAMSUNG ELECTRONICS CO., LTD.; AND
`SAMSUNG ELECTRONICS AMERICA, INC.
`Petitioner
`
`v.
`
`IMAGE PROCESSING TECHNOLOGIES, LLC
`Patent Owner
`
`____________________
`
`Patent No. 6,717,518
`____________________
`
`PETITION FOR INTER PARTES REVIEW
`OF U.S. PATENT NO. 6,717,518
`
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`I.
`
`TABLE OF CONTENTS
`
`Contents
`INTRODUCTION ........................................................................................... 1
`
`II. MANDATORY NOTICES UNDER 37 C.F.R. § 42.8 ................................... 1
`
`III.
`
`PAYMENT OF FEES UNDER 37 C.F.R. § 42.15(a) .................................... 2
`
`IV. GROUNDS FOR STANDING ........................................................................ 3
`
`V.
`
`PRECISE RELIEF REQUESTED .................................................................. 3
`
`VI. LEGAL STANDARDS ................................................................................... 3
`
`A.
`
`B.
`
`Claim Construction ............................................................................... 3
`
`Level of Ordinary Skill In The Art ....................................................... 4
`
`VII. OVERVIEW OF THE ’518 PATENT ............................................................ 4
`
`VIII. DETAILED EXPLANATION OF GROUNDS ............................................ 10
`
`A. Overview Of The Prior Art References .............................................. 10
`
`1. Martin Eriksson et al., Eye Tracking for Detection of
`Driver Fatigue, IEEE Conference on Intelligent
`Transportation Systems (Nov. 1997) (“Eriksson”) (Ex.
`1005) ......................................................................................... 10
`
`2.
`
`3.
`
`4.
`
`Luigi Stringa, Eyes Detection For Face Recognition,
`Applied Artificial Intelligence (1993) (“Stringa”) (Ex.
`1006) ......................................................................................... 15
`
`U.S. Patent No. 5,805,720, Facial Image Processing
`System (Filed Mar. 11, 1996) (“Suenaga”) (Ex. 1007) ............. 18
`
`U.S. Patent No. 5,008,946, System For Recognizing
`Image (Filed Sept. 9, 1988) (“Ando”) (Ex. 1009) .................... 21
`
`IX. SPECIFIC EXPLANATION OF GROUNDS FOR INVALIDITY ............. 26
`
`i
`
`
`
`Petition for Inter Partes Review
`Patent No. 7,650,015
`A. Ground 1: Eriksson In View Of Stringa Renders Obvious Claim
`39 ......................................................................................................... 26
`
`1.
`
`2.
`
`3.
`
`Reasons To Combine Eriksson And Stringa ............................ 26
`
`Claim 39 .................................................................................... 29
`
`Eriksson and Stringa Are Not Cumulative ............................... 40
`
`A. Ground 2: Ando In View Of Suenaga Renders Obvious Claim
`39 ......................................................................................................... 41
`
`1.
`
`2.
`
`3.
`
`Reasons To Combine Ando and Suenaga ................................. 41
`
`Claim 39 .................................................................................... 44
`
`Ando and Suenaga Are Not Cumulative ................................... 56
`
`B.
`
`Ground 3: Ando In View Of Stringa Renders Obvious Claim 39 ...... 56
`
`1.
`
`2.
`
`3.
`
`Reasons To Combine Ando and Stringa ................................... 56
`
`Claim 39 .................................................................................... 59
`
`Ando and Stringa Are Not Cumulative ..................................... 69
`
`X.
`
`CONCLUSION .............................................................................................. 70
`
`
`
`
`
`
`
`-ii-
`
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`LIST OF EXHIBITS1
`
`U.S. Patent No. 6,717,518 (“the ’518 Patent”)
`Declaration of Dr. John C. Hart
`Curriculum Vitae for Dr. John C. Hart
`Prosecution File History of U.S. Patent No. 6,717,518
`Martin Eriksson et al., Eye Tracking For Detection Of Driver
`Fatigue, IEEE Conference on Intelligent Transportation
`Systems (Nov. 1997) (“Eriksson”)
`Luigi Stringa, Eyes Detection For Face Recognition, Applied
`Artificial Intelligence (1993) (“Stringa”)
`U.S. Patent No. 5,805,720, Facial Image Processing System
`(Filed Mar. 11, 1996) (“Suenaga”)
`U.S. Patent No. 5,293,427, Eye Position Detecting System and
`Method Therefor (Filed Dec. 11, 1991) (“Ueno”)
`U.S. Patent No. 5,008,946, System For Recognizing Image
`(Filed Sept. 9, 1988) (“Ando”)
`Declaration of William Garrity from U.C. Davis Regarding
`Stringa
`Declaration of Dr. Umit Ozguner Regarding Eriksson
`
`1001
`1002
`1003
`1004
`1005
`
`1006
`
`1007
`
`1008
`
`1009
`
`1010
`
`1011
`
`
`
` 1
`
` Citations to non-patent publications are to the original page numbers of the
`
`publication, and citations to U.S. patents are to column:line number of the patents.
`
`iii
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`I.
`
`INTRODUCTION
`
`Samsung Electronics Co., Ltd. and Samsung Electronics America, Inc.
`
`(collectively, “Petitioner”) request inter partes review (“IPR”) of Claim 39 of U.S.
`
`Patent No. 6,717,518 (“the ’518 Patent”) (Ex. 1001), which Petitioner understands
`
`to be currently assigned to Image Processing Technologies, LLC (“Patent
`
`Owner”). This Petition presents three non-cumulative grounds of invalidity that
`
`the U.S. Patent and Trademark Office (“PTO”) did not consider during
`
`prosecution. These grounds are each likely to prevail, and this Petition,
`
`accordingly, should be granted on all grounds and the challenged claim should be
`
`cancelled.
`
`II. MANDATORY NOTICES UNDER 37 C.F.R. § 42.8
`Real Parties-in-Interest: Petitioner identifies the following real parties-in-
`
`interest: Samsung Electronics Co., Ltd. and Samsung Electronics America, Inc.
`
`Related Matters: Patent Owner has asserted the ’518 Patent against
`
`Petitioner in Image Processing Technologies LLC v. Samsung Elecs. Co., No.
`
`2:16-cv-00505-JRG (E.D. Tex.). Patent Owner has also asserted U.S. Patent Nos.
`
`6,959,293; 8,805,001; 8,983,134; 7,650,015; and 8,989,445 in the related action.
`
`Petitioner is concurrently filing additional IPR petitions for several of these
`
`asserted patents, and has previously filed the following IPR petitions:
`
`• IPR2017-00355 against U.S. Patent 7,650,015
`
`1
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`• PR2017-00357 against U.S. Patent 8,989,445
`
`• IPR2017-00336 against U.S. Patent 6,959,293
`
`• IPR2017-00347 against U.S. Patent 8,805,001
`
`• IPR2017-00353 against U.S. Patent 8,983,134
`
`Lead and Back-Up Counsel:
`
`• Lead Counsel: John Kappos (Reg. No. 37,861), O'Melveny & Myers
`
`LLP, 610 Newport Center Drive, 17th Floor, Newport Beach,
`
`California 92660. (Telephone: 949-823-6900; Fax: 949-823-6994;
`
`Email: jkappos@omm.com.)
`
`• Backup Counsel: Nicholas J. Whilt (Reg. No. 72,081), Brian M. Cook
`
`(Reg. No. 59,356), O'Melveny & Myers LLP, 400 S. Hope Street, Los
`
`Angeles, CA 90071. (Telephone: 213-430-6000; Fax: 213-430-6407;
`
`Email: nwhilt@omm.com, bcook@omm.com.)
`
`Service Information: Samsung consents to electronic service by email to
`
`IPTSAMSUNGOMM@OMM.COM. Please address all postal and hand-delivery
`
`correspondence to lead counsel at O’Melveny & Myers LLP, 610 Newport Center
`
`Drive, 17th Floor, Newport Beach, California 92660, with courtesy copies to the
`
`email address identified above.
`
`III. PAYMENT OF FEES UNDER 37 C.F.R. § 42.15(a)
`The Office is authorized to charge an amount in the sum of $23,000 to
`
`2
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`Deposit Account No. 50-2862 for the fee set forth in 37 CFR § 42.15(a), and any
`
`additional fees that might be due in connection with this Petition.
`
`IV. GROUNDS FOR STANDING
`Petitioner certifies that the ’518 Patent is available for IPR and Petitioner is
`
`not barred or estopped from requesting IPR on the grounds identified herein.
`
`V.
`
`PRECISE RELIEF REQUESTED
`
`Petitioner respectfully requests review and cancellation of Claim 39 of the
`
`’518 Patent based on three grounds:
`
`• Ground 1: Claim 39 is obvious under 35 U.S.C. § 103(a) over
`
`Eriksson in view of Stringa.
`
`• Ground 2: Claim 39 is obvious under 35 U.S.C. § 103(a) over Ando
`
`in view of Suenaga.
`
`• Ground 3: Claim 39 is obvious under 35 U.S.C. § 103(a) over Ando
`
`in view of Stringa.
`
`VI. LEGAL STANDARDS
`A. Claim Construction
`In an inter partes review, “[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). The ’518 patent will not expire before a final
`
`written decision issues, and its claims should be given their broadest reasonable
`
`3
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`interpretation.2 Petitioner submits that for purposes of this Petition, no special
`
`definition applies to any term of Claim 39, and the terms should be interpreted
`
`according to their ordinary and customary meaning. Ex. 1002, Hart Decl. ¶ 47.
`
`Level of Ordinary Skill In The Art
`
`B.
`One of ordinary skill in the art the time of the alleged invention of the ’518
`
`Patent would have had either (1) a Master’s Degree in Electrical Engineering or
`
`Computer Science or the equivalent plus at least a year of experience in the field of
`
`image processing, image recognition, machine vision, or a related field or (2) a
`
`Bachelor’s Degree in Electrical Engineering or Computer Science or the equivalent
`
`plus at least three years of experience in the field of image processing, image
`
`recognition, machine vision, or a related field. Additional education could
`
`substitute for work experience and vice versa. Ex. 1002, ¶¶ 44–46.
`
`VII. OVERVIEW OF THE ’518 PATENT
`The ’518 Patent purports to disclose an application for the inventor’s
`
`
` Because the claim construction standard in this proceeding differs from the
`
` 2
`
`standard applicable to a district court litigation, see In re Am. Acad. of Sci. Tech
`
`Ctr., 367 F.3d 1359, 1364, 1369 (Fed. Cir. 2004), Petitioner expressly reserves the
`
`right to argue in litigation a different construction for any term recited by the
`
`claims of the ’518 Patent.
`
`4
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`previously patented “generic image processing system . . . ” (“GIPS”). Ex. 1001 at
`
`2:1–5. Specifically, the ’518 Patent proposes applying GIPS to “detect the
`
`drowsiness of a person.” Id. at 2:28–29. The patent explains that drowsiness
`
`detection addresses the problem that “a significant number of highway accidents
`
`result from drivers becoming drowsy or falling asleep . . . .” Id. at 1:12–17.
`
`Drowsiness can be detected by the duration of blinks (i.e., longer blinks occur
`
`when a driver becomes drowsy). Id. at 1:18–24. Thus, the Patent proposes
`
`mounting a video camera in a car and detecting blink rates using GIPS. Id. at
`
`6:28–56.
`
`For example, when the driver enters the vehicle, GIPS could detect the
`
`driver by looking for pixels that are “moving in a lateral direction away from the
`
`driver’s door” and that have the “hue characteristics of skin.” Id. at 25:24–39; Ex.
`
`1002, ¶ 33. Knowing a driver is present, GIPS then “detects the face of the driver
`
`in the video signal and eliminates from further processing those superfluous
`
`portions of the video signal above, below, and to the right and left of the head of
`
`the driver.” Ex. 1001 at 26:16–22. Specifically, the head is detected by looking
`
`for pixels with “selected characteristics” such as pixels that appear to be moving or
`
`to have a skin color. Id. at 26:21–45. These pixels could then be loaded into
`
`several histograms (324x and 324y), as shown below:
`
`5
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`Ex. 1001, Fig. 24
`
`
`
`Thus, for example, the head (in the region V) could be detected in the figure
`
`above by looking for peaks in the histogram, which can indicate the edge of the
`
`face. Id. at 26:49–65. Alternatively, GIPS could search for groups of pixels with
`
`“low luminance levels” to identify “nostrils.” Id. at 29:18–29.
`
`GIPS can then ignore the area in the frame outside of the face, and only
`
`continue with analyzing the face (V), which would be in the region Z bounded by
`
`Ya, Yb, Xc, and Xd in Figure 25, below. Id. at 26:66–27:10.
`
`6
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`The patent calls the exclusion of the hashmarked background area in the
`
`frame “masking.” Id. at 26:66–27:1.
`
`Next, GIPS “uses the usual anthropomorphic ratio between the zone of the
`
`eyes and the entire face for a human being” to obtain a mask for the eyes of the
`
`driver. Id. at 27:33–38. Use of an anthropomorphic model is explained to refer to
`
`using a “facial characteristic, e.g., the nose, ears, eyebrows, mouth, etc., and
`
`combinations thereof” or “the outline of the head of the driver” as a “starting point
`
`for locating the eyes.” Id. at 29:43–56; Ex. 1002, ¶ 37. The patent explains that
`
`the sub-area can also be “set using an anthropomorphic model, wherein the spatial
`
`relationship between the eyes and nose of humans is known.” Ex. 1001 at 30:43–
`
`45. Thus, using the anthropomorphic model, the patent proposes deriving the sub
`
`area Zʹ from the larger face area Z, as indicated below. Id. at 27:31–38; Ex. 1002,
`
`¶ 37.
`
`7
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`
`
`For example, the patent explains that the “nostrils 272” can be used to
`
`identify a “search box 276” around the “eye 274 of the driver,” as shown in Figure
`
`32, using “an anthropomorphic model.” Ex. 1001 at 30:40–45.
`
`
`
`Ex. 1001, Fig. 32
`
`
`
`Having reduced the area for processing to a smaller region that contains the
`
`eye, GIPS can then check for blinks by “analyzing the pixels within the area Zʹ to
`
`identify” blinking. Id. at 27:54–55, 31:3–9. The Patent proposes a variety of
`
`methods to identify blinking, such as (1) “analyzing the shape of the eye
`
`shadowing to identify shapes corresponding to openings and closings of the eye,”
`
`(id. at 4:25–33, 31:10–17), (2) analyzing pixels in the eye area with “high speed
`
`8
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`vertical movement” with “the hue of skin” (id. at 27:56–57), or (3) analyzing
`
`pixels in the eye area that lack “the hue of skin” (id. at 27:62–65). Ex. 1002, ¶ 39.
`
`Figure 27, below, shows the use of histograms to analyze the pixels in the eye
`
`area—peaks can indicate whether the eye is open or closed. Ex. 1001 at 28:47–51.
`
`
`
`
`
`The patent proposes that these histograms can be created for each frame, and
`
`changes in the histograms over time can be analyzed to determine blink rates. Id.
`
`at 28:32–29:10; Ex. 1002, ¶ 40. For example, Figure 33 shows the histograms for
`
`an open eye (featuring large peaks), and Figure 34 shows the histograms for a
`
`closed eye (featuring small peaks):
`
`
`
`
`
`Ex. 1001, Fig. 33
`
`
`
`
`
`
`
`
`
`Ex. 1001, Fig. 34
`
`
`
`9
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`The patent also proposes searching for “characteristics indicative of an eye
`
`
`
`present in the search box,” such as “a moving eyelid, a pupil, iris or cornea, a
`
`shape corresponding to an eye, a shadow corresponding to an eye, or any other
`
`indicia indicative of an eye.” Id. at 30:56–59; Ex. 1002, ¶ 41. Thus, for example,
`
`Figure 36 “shows a sample histogram of a pupil 432,” formed by “detect[ing]
`
`pixels with very low luminance levels and high gloss that are characteristic of a
`
`pupil.” Ex. 1001 at 30:61–64.
`
`
`
`Ex. 1001, Fig. 36
`
`VIII. DETAILED EXPLANATION OF GROUNDS
`A. Overview Of The Prior Art References
`1. Martin Eriksson et al., Eye Tracking for Detection of Driver
`Fatigue, IEEE Conference on Intelligent Transportation
`Systems (Nov. 1997) (“Eriksson”) (Ex. 1005)
`
`In the 1990s, the detection of driver fatigue was the subject of government
`
`funding by institutions such as the Minnesota Department of Transportation, the
`
`National Science Foundation, and the Center for Transportation Studies. See, e.g.,
`
`Ex. 1005 at 319. Pursuant to that funding, Martin Eriksson and Professor Nikolaos
`10
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`Papanikolopoulos at the University of Minnesota developed a system to detect
`
`driver fatigue that is very similar to the process of Claim 39 of the ’518 Patent.
`
`IEEE made Eriksson publicly available at an IEEE conference from November 9–
`
`12, 1997. See Ex. 1011 at ¶¶ 2–4. Thus, Eriksson is prior art at least under pre-
`
`AIA 35 U.S.C. § 102(a) and (b) and is a statutory bar under pre-AIA 35 U.S.C.
`
`§ 119.
`
`Eriksson “describe[s] a system that locates and tracks the eyes of a driver”
`
`for the “purpose of . . . detect[ing] driver fatigue.” Ex. 1005 at 314. Eriksson
`
`proposes mounting “a small camera inside the car” to “monitor the face of the
`
`driver and look for eye movements which indicate that the driver is no longer in
`
`condition to drive.” Id. at 314. Eriksson notes that “[a]s the driver becomes more
`
`fatigued, we expect the eye blinks to last longer.” Id. at 317. Thus, Eriksson
`
`proposes a system for detecting the driver’s pupil—when the pupil is detected, the
`
`eye is open, and when the pupil is not detected, the eye is closed. Id. at 318.
`
`Eriksson determines the location of the eyes in four steps. Id. at 315. The
`
`first step is “localization of the face.” Id. Eriksson explains that the face is
`
`localized using a “symmetry histogram.” Id.
`
`11
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`
`
`Eriksson calculates a “symmetry-value” for each pixel-column in order to
`
`find the center of the face. Id. at 316. The pixel column with the lowest symmetry
`
`value will be the center of the face. Id.; Ex. 1002, ¶ 50. Then, having identified
`
`the center of the face, Eriksson narrows the search area to a smaller area that
`
`includes the eyes: “the search-space is . . . limited to the area around this line,
`
`which reduces the probability of having distracting features in the background.”
`
`Ex. 1005 at 316.
`
`
`
`The second step in localizing the face is computing the vertical location of
`
`the eyes. Id. To do this, Eriksson creates a gradient histogram of the sub-area of
`
`the image identified in the first step, as illustrated in Figure 2:
`
`12
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`
`
`Eriksson “consider[s] the best three peaks in” the histogram (which in the
`
`above example appear to correspond to the eyes, the nose, and the mouth) as
`
`potential vertical locations for the eyes. Ex. 1005 at 316.
`
`
`
`The third step in localizing the eyes is finding “the exact location of the
`
`eyes.” Id. at 316. Having limited the search for the eyes to the horizontal region
`
`determined in the first step, and the three possible vertical locations determined in
`
`the second step, Eriksson finds the eyes by searching for “intensity-valleys” in the
`
`image and also using “general constraints, such [as] that both eyes must be located
`
`‘fairly close’ to the center of the face.” Id.
`
`
`
`The fourth step in localizing the eyes is estimating the position of the iris.
`
`Eriksson uses an “eye-template,” shown below, that, when laid over the picture,
`
`indicates a good match if there are “many dark pixels in the area inside the inner
`
`circle, and many bright pixels in the area between the two circles.” Id. at 316–17.
`
`When a match occurs, Eriksson knows “the inner circle is centered on the iris and
`
`the outside circle covers the sclera.” Id. at 317.
`
`13
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`
`
`Having found the eye, Eriksson next generates a horizontal intensity
`
`histogram across the pupil. Id. at 318. Eriksson notes that the pupil and iris are
`
`dark and the sclera is light. Id. Thus, the histogram of an open eye is markedly
`
`different from the histogram of a closed eye:
`
`
`
`Finally, having found the iris, pupil, and sclera, and having determined
`
`whether the eye is open or closed in each frame, Eriksson is able to measure blink
`
`rates over time and detect drowsy drivers. Id. at 318.
`
`
`
`14
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`Luigi Stringa, Eyes Detection For Face Recognition, Applied
`Artificial Intelligence (1993) (“Stringa”) (Ex. 1006)
`
`2.
`
`Stringa is a printed publication published and made publicly available in
`
`1993. See Ex. 1010. Thus, Stringa is prior art to the ’518 Patent under at least 35
`
`U.S.C. § 102(a) and (b).
`
`Stringa discloses an image processing normalization algorithm for
`
`improving previously developed algorithms for face detection. Ex. 1006 at 365.
`
`Stringa explains that for face recognition systems, sometimes captured faces are
`
`not looking “straight into the camera” and thus “some adjustment and
`
`normalization is necessary before the system can proceed to the recognition step.”
`
`Id. at 366. As part of this normalization procedure, Stringa discloses detecting the
`
`pupils of the face in a manner similar to the ’518 Patent, especially with respect to
`
`Claim 39’s use of an anthropometric model.
`
`Stringa explains that its approach to “locating the position of the eyes” is
`
`“based on the exploitation of (a priori) anthropometric information combined with
`
`the analysis of suitable grey-level distributions, allowing direct localization of both
`
`eyes.” Ex. 1006 at 369. Stringa explains that
`
`there exists a sort of ‘grammer’ of facial structures that provides some
`
`very basic a priori information used in the recognition of faces. Every
`
`human face presents a reasonable symmetry, and the knowledge of the
`
`relative position of the main facial features (nose between eyes and
`15
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`over mouth, etc.) proves very useful to discriminate among various
`
`hypotheses. These guidelines can be derived from anthropometric
`
`data corresponding to an average face and refined through the analysis
`
`of real faces. Some typical examples . . . are:
`
`• the eyes are located halfway between the top of the head and
`
`the bottom of the chin;
`
`• the eyes are about one eye width apart;
`
`• the bottom of the nose is halfway between the eyebrows and the
`
`chin; . . . .
`
`Ex. 1006 at 369.
`
`
`
`Stringa’s eye localization algorithm first detects the line that connects the
`
`eyes, then the side limits of the face and the nose axis. Id. at 370. To obtain the
`
`pupil location, Stringa first uses “the approximate location of the eye-connecting
`
`line, of the face sides, and of the nose axis” to estimate “the expectation zones of
`
`the two eyes . . . with reasonable accuracy.” Id. at 376. Stringa illustrates “the
`
`expectation zones for the two eyes” in Figure 9:
`
`16
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`In the expectation zones for the two eyes, “the search of the pupil is based
`
`on the analysis of the horizontal grey-level distribution,” (i.e., a histogram). Id. at
`
`377; Ex. 1002, ¶ 60. Stringa uses the histogram and further mathematical
`
`calculations to produce a graph that identifies pupil (Ex. 1006 at 377):
`
`
`
`17
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`U.S. Patent No. 5,805,720, Facial Image Processing System
`(Filed Mar. 11, 1996) (“Suenaga”) (Ex. 1007)
`
`3.
`
`Suenaga is a U.S. Patent filed on March 11, 1996 and issued on Sept. 28,
`
`1998. Ex. 1007. Thus, Suenaga is prior art to the ’518 Patent under at least 35
`
`U.S.C. § 102(a), (b) & (e).
`
`Suenaga discloses a “facial image processing system for detecting . . . a
`
`dozing or drowsy condition of an automobile driver . . . from the opened and
`
`closed conditions of his eyes.” Ex. 1007 at 1:6–10. Suenaga uses a video camera
`
`to obtain images of a face. Id. at 2:44–49; 6:25–35.
`
`Suenaga discloses many embodiments. Embodiment 31 explains that boxes
`
`11, 12, and 13 (in Fig. 60, below) in the flowchart for Embodiment 31 are the same
`
`as those steps in Embodiment 1. Id. at 23:19–21. Embodiment 1 explains that in
`
`boxes 11, 12, and 13, Suenaga converts the image from the camera into a binary
`
`image (i.e., each pixel is assigned to be a one or a zero). Id. at 6:41–51.
`
`
`
`Ex. 1007, Fig. 60
`
`18
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`For box 15A, the evaluation function calculation means “first finds the
`
`
`
`barycenter or centroid 31 from the average of the coordinates of black pixels in a
`
`binary image 30” (id. at 23:21–24):
`
`Ex. 1007, Fig. 61
`
`
`
`
`
`Next, “rectangular areas existing in the predetermined ranges in the X-
`
`direction on the left and right sides of this barycenter or centroid 31 are set as eye
`
`presence areas 32.” Id. at 23:24–27. Then, “in the eye presence area 32, X-
`
`histograms 33 (namely, 33a and 33b) are generated.” Id. at 23:27–29. Next,
`
`“zonal regions are set on the basis of the X-histograms. Furthermore, Y-
`
`19
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`histograms 34 (namely, 34a and 34b) . . . are produced.” Id. at 23:29–34.
`
`Additionally, “hatched candidate areas 35 (namely 35a and 35b) for an eye
`
`presence area are extracted.” Id. at 23:34–35.
`
`
`
`Next, Embodiment 31 calculates the evaluation function as described in the
`
`previous embodiments. Id. at 23:36–40. In Embodiment 1, the “the evaluation
`
`function calculation means 15 calculates an evaluation function, which”
`
`determines the “shape of the eye.” Id. at 6:61–65.
`
`
`
`Figure 2, below, illustrates how Suenaga determines whether the eye is open
`
`or closed in Embodiments 1 and 31. Ex. 1007 at 7:4–24, 23:52–54. Suenaga
`
`examines the shape of the eye by analyzing the histograms of the eye pixels. Id. at
`
`7:4–65.
`
`
`
`Ex. 1007, Figure 2
`
`20
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`Suenaga calculates a “value K” based on the lines 9 and 10 calculated from
`
`
`
`the shape of the histogram curve. Id. at 7:28–65. Suenaga determines whether the
`
`eye is open or closed based on the relationship of K to a threshold value, KB. Id.
`
`As shown in Figure 2, this evaluation is performed over time. Id. at 7:4–24 (“This
`
`diagram illustrates the relation among the lapse of time (from a moment TA, at
`
`which the eye is opened, to another moment TC, at which the eye is closed . . . ).”).
`
`4.
`
`U.S. Patent No. 5,008,946, System For Recognizing Image
`(Filed Sept. 9, 1988) (“Ando”) (Ex. 1009)
`
`Ando was filed in September, 1988 and issued in April, 1991. Thus, Ando
`
`is prior art to the ’518 Patent under at least 35 U.S.C. § 102(a), (b) & (e).
`
`Ando discloses a system for detecting certain portions of an image, such as
`
`“the driver’s eyes and mouth.” Ex. 1009 at 2:1–4. The system uses information
`
`about the driver’s eyes, such as the position of the eyes and whether they are open
`
`or closed, to allow the driver to “control electrical devices,” such as the windows
`
`and radio, “in a noncontact manner.” Id. at 2:18–20, Fig. 1a. Ando’s hands-free
`
`control system increases the safety and comfort of driving. Id. at 2:58–59. The
`
`system uses a video camera 3 mounted on the dashboard (id. at 6:60–7:3):
`
`21
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`
`
`Ex. 1009, Fig. 1b (showing video camera 3 on dashboard)
`
`Ando includes “a window setting means for setting a region narrower than
`
`the image produced by the camera means.” Id. at 2:25–41. “Once the position of
`
`the certain portions, i.e., the eyes and mouth, are detected, the scan made to detect
`
`the eyes and mouth is limited to the narrower region [] so they can be detected
`
`quickly.” Id. Ando “can detect the driver’s head, face, and pupils with high
`
`accuracy.” Id. at 3:30–31.
`
`Ando’s algorithm has two main phases, described in more detail below. The
`
`first phase operates on the first frame and determines some parameters for tracking
`
`the pupils in later frames. Id. at 35:14–36:31. The second phase operates on
`
`subsequent frames and tracks the pupils using the information calculated from the
`
`first frame. Id. at 36:32–44. If detection fails in the subsequent frames, the first
`
`phase is repeated. Id. at 36:44–51.
`
`22
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`
`a)
`In phase one, Ando identifies likely locations for the head HDD, then the
`
`Phase One
`
`forehead BRD, the right eye RED, the left eye LED, the eyebrows 35, the mouth
`
`MOD, and the nose NOD. Id. at Fig. 2. Ando uses histograms to calculate
`
`thresholds for distinguishing those face elements from other elements in the frame.
`
`Id. at Figs. 5b (calculating threshold for head detection), 7b (calculating threshold
`
`for forehead detection), 8b (calculating pupil threshold 25). Ando also identifies
`
`windows, or portions of the frame, for where those elements are located within the
`
`frame. Id. at Figs. 2 (identifying head and forehead windows in 34 & 39), 8b
`
`(identifying pupil region Sd 113).
`
`For example, to find the head, Ando uses a histogram of the gray level of
`
`each pixel. Id. at 14:23–16:32, Figs. 5a–5c. As part of the head detection process,
`
`Ando calculates the width of the head AW. Id. at 16:7–13. Similarly, Ando finds
`
`the forehead using another gray level histogram. Id. at 16:44–57, Figs. 7a–7c. For
`
`the forehead, Ando calculates the upper boundary HTY, the right end boundary
`
`HRX, the left end boundary HLX, and the width HW. Id. at 17:52–54, Fig. 13d.
`
`Ando also detects the right and left eyes. Id. at 17:63–21:53. As part of eye
`
`detection, Ando defines a portion of the image Sd which is calculated based on the
`
`forehead boundaries (id. at 18:11–14) and the expected position of the eye using
`
`23
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`known ratios for a human face (i.e., an anthropomorphic model), as shown in
`
`Figure 13d:
`
`Ex. 1009, Fig. 13d (showing portion Sd)
`
`
`
`Having defined the portion of the image Sd, Ando calculates a gray level
`
`histogram for that portion in order to identify the pupils specifically. Id. at 18:15–
`
`20:52, Figs. 8a–8d. Then, Ando detects the mouth and the nose using similar
`
`processes. Id. at Fig. 2 (MOD and NOD), 22:8–62.
`
`Having identified likely locations for the face elements, Ando next conducts
`
`a face verification process FAD to verify whether the relative locations of those
`
`face elements imply they are indeed part of a face, and are thus the elements they
`
`appear to be. Specifically, Ando uses a “similarity degree-detecting circuit,” to
`
`check whether the positions and locations of those elements indicate that they are
`
`similar to “reference values” for a face. Id. at 12:20–26, 4:66–44; 22:63–27:35
`
`24
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`39:49–41:14, Figs. 9a–9d, Table 1. As part of this process, Ando uses an
`
`anthropomorphic model created by measuring the faces of his acquaintances and
`
`himself. Id. at 22:67–23:32.
`
`Ando’s “similarity degree-detecting circuit” measures “the degrees of
`
`similarity of the detected elements to the elements of [a] reference image.” Id. at
`
`5:19–29. The reference image consists of “statistical values” describing the
`
`“shapes and relative positions” of “ordinary persons.” Id. at 39:58–63. If the
`
`image has face elements that are similar in shape and position to those elements in
`
`the face of an ordinary person, Ando determines that the image contains a face and
`
`that the identified face elements are located where they were calculated to be. Id.
`
`at 12:23–31. Upon positive verification, Ando then proceeds to phase 2. Id. at
`
`12:27–32.
`
`b)
`In phase two (summarized in boxes 40–54 in Figure 2), Ando retrieves a
`
`Phase Two
`
`new frame from the camera and uses the thresholds and windows calculated in
`
`phase one to search for the pupils 44 and the mouth 51. Id. at 12:31–33. After
`
`finding the pupils, Ando checks whether the eyes are open or closed and which
`
`direction the pupil is looking 49. Id. at 12:33–35. Specifically, Ando is “equipped
`
`with a state change-detecting means for detecting the states of the eyes and mouth
`
`at successive instants in time to detect the changes in the states.” Id. at 2:42–46.
`
`25
`
`
`
`
`Petition for Inter Partes Review
`Patent No. 6,717,518
`Thus, “when the states of the monitored eyes and mo