MORPHO

Patent Trials and Appeals Board

Feb 2, 2022

2020005641 (P.T.A.B. Feb. 2, 2022)

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. 14/437,803 04/22/2015 Cedric THUILLIER 4005/0563PUS1 8655 60601 7590 02/02/2022 Muncy, Geissler, Olds & Lowe, P.C. 4000 Legato Road Suite 310 Fairfax, VA 22033 EXAMINER ZHAO, YU ART UNIT PAPER NUMBER 2169 NOTIFICATION DATE DELIVERY MODE 02/02/2022 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): MAILROOM@MG-IP.COM PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE ________________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ________________ Ex parte CEDRIC THUILLIER and JEROME BAYON DE NOYER ________________ Appeal 2020-005641 Application 14/437,803 Technology Center 2100 ________________ Before JASON V. MORGAN, JEREMY J. CURCURI, and HUNG H. BUI, Administrative Patent Judges. MORGAN, Administrative Patent Judge. DECISION ON APPEAL STATEMENT OF THE CASE Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the Examiner’s decision to reject claims 1-7. We have jurisdiction under 35 U.S.C. § 6(b). We reverse. 1 “Appellant” refers to “applicant” as defined in 37 C.F.R. § 1.42. Appellant identifies IDEMIA IDENTITY & SECURITY as the real party in interest. Appeal Br. 2. Appeal 2020-005641 Application 14/437,803 2 SUMMARY OF THE DISCLOSURE Appellant’s claimed subject matter relates to identifying a subject person by comparing at least biometric data of the subject person with biometric data of reference people previously recorded in a database. Abstract. In performing this comparison, either a first algorithm or a second algorithm “is selected on the basis of at least one selection criterion specific to each reference person.” Id. REFERENCES The Examiner relies on the following references (only the first named inventor of each reference is listed): Name Reference Date Chan US 2006/0171571 A1 Aug. 3, 2006 Yamada US 2010/0148922 A1 June 17, 2010 Robinson US 2013/0090942 A1 Apr. 11, 2013 Tony Mansfield, The Application of Quality Scores in Biometric Recognition, NIST Biometric Quality Workshop II (Nov. 7, 2007) (“Mansfield”) REJECTIONS The Examiner rejects claims 1-7 as follows: Claims Rejected 35 U.S.C. § Reference(s)/Basis Citation 1-6 103(a) Chan, Mansfield, Yamada Final Act. 14-27 7 103(a) Chan, Mansfield, Yamada, Robinson Final Act. 27-29 Appeal 2020-005641 Application 14/437,803 3 ILLUSTRATIVE CLAIM Claim 1, which is illustrative with respect to claims 1-7, is reproduced below (key disputed recitations highlighted and bracketed numerals added for clarity). 1. A method of identifying a subject person, comprising: [1] recording in advance, in a database, biometric data of a type of reference persons; [2] determining, for each reference person, a quality level of the biometric data of the type; [3] obtaining current biometric data of the type of the subject person by using acquisition means comprising at least a sensor and an analog to digital converter; [4] for each reference person, dynamically selecting, by the computer, one of a first algorithm and a second algorithm based on a comparison of the quality level of the biometric data of the type for the reference person and a predetermined threshold value, [5] the second algorithm requiring longer execution time but providing a more reliable result than the first algorithm, [6] the first algorithm being selected if the quality level of the biometric data of the type for the reference person is higher than the predetermined threshold value, or if uniqueness of the biometric data of the type for the reference person is high concerning the biometric data of the type, the second algorithm being selected if the quality level is lower than the predetermined threshold value, or if the uniqueness of the biometric data of the type for the reference people is low concerning the biometric data of the type; [7] for each reference person, comparing, by the computer, the biometric data of the type of the subject person with biometric data of the type of the reference person by using the selected one of the first algorithm and the second algorithm; and Appeal 2020-005641 Application 14/437,803 4 [8] generating a result of the comparison of the biometric data of the type of the subject person and the biometric data of the type of the reference person. Appeal Br. 16 (Claims App.). ANALYSIS The Examiner relies on the combination of (1) Chan’s use of multiple biometrics for authentication, (2) Mansfield’s use of quality scores in biometric recognition, and (3) Yamada’s selection of high-speed or precise verification algorithms to teach or suggest the recitations of claim 1. In particular, the Examiner finds that Chan’s matching of sampled fingerprint information (i.e., a subject person’s biometric data) to fingerprint samples in a database (i.e., reference person’s biometric data) to produce a fused score (i.e., to generate a result of the comparison) teaches or suggest recitations [1]-[3], [7], and [8]. Final Act. 14-18 (citing, e.g., Chan ¶¶ 14, 26, 27, 40- 42); Ans. 4-15 (citing, e.g., Chan Abstract, ¶¶ 17, 19, 25). The Examiner then relies on Mansfield’s use of a different algorithm if the quality score for biometric data is low to teach or suggest recitations [4] and [6]. Final Act. 19-20 (citing Mansfield 3); Ans. 17-19. The Examiner concludes it would have been obvious to modify Chan using Mansfield’s teachings and suggestions to improve performance if quality is poor. Final Act. 20 (citing Mansfield 3). The Examiner also finds that Yamada’s distinction between a high- speed and a precise verification process teaches or suggests recitation [5]. Final Act. 20-21 (citing, e.g., Yamada ¶¶ 28, 30, 88). The Examiner further finds that Yamada’s selection of either a high-speed or a precise verification process provides alternative teachings and suggestions with respect to recitations [4] and [6]. Ans. 15-17 (citing Yamada ¶¶ 4, 278-79). The Appeal 2020-005641 Application 14/437,803 5 Examiner concludes that modifying Chan using the teachings and suggestions of Yamada would have been obvious “because it would provide Chan’s modified method with the enhanced capability of ‘. . . requires a larger amount of computation than the high-speed verification process but is with a higher precision than the high-speed verification process.’” Final Act. 21 (some emphasis omitted).2 Appellant contends the Examiner erred because Mansfield’s disclosure to “do something different if quality is poor” such as “use [a] different algorithm” does not teach or suggest recitation [4], for each reference person, dynamically selecting, by the computer, one of a first algorithm and a second algorithm based on a comparison of the quality level of the biometric data of the type for the reference person and a predetermined threshold value, [6] the first algorithm being selected if the quality level of the biometric data of the type for the reference person is higher than the predetermined threshold value and the second algorithm being selected if the quality level is lower than the predetermined threshold value. Appeal Br. 12 (quoting Mansfield 3). In particular, Appellant argues this sentence “can be interpreted in many different ways,” including, e.g., modifying an algorithm “depending on the subject person (and not on each reference person).” Id. In response, the Examiner takes the position that Mansfield teaches the use of quality scores during enrollment to indicates that such quality 2 The precise quotation does not exist in Chan’s paragraph 81, as the Examiner purports, but the substance of the misquoted language is present in Chan’s paragraphs 79 and 80. Appeal 2020-005641 Application 14/437,803 6 scores refer to biometric data of a reference person. Ans. 20 (citing Mansfield 3). The Examiner’s reliance on Mansfield, however, is not persuasive. Recitations [4] and [6] recite dynamically selecting a first algorithm or a second algorithm based on whether a determined quality level for biometric data exceeds a predetermined threshold. The disclosure of using a different algorithm at best suggests using a first algorithm and then using a second algorithm if one or more quality metrics from the first algorithm are poor. This accords with the other remedies of Mansfield, which include retaking an image, taking an additional image, and taking remedial correction of specific problems such as correcting a pose. Mansfield 3. That is, all the actions in the cited portion of Mansfield used “[t]o improve performance if quality is poor” (id.) relate to doing something additional when quality is deemed poor, rather than making a selection (e.g., between two algorithms) based on quality. Thus, even if Mansfield uses a different algorithm during enrollment of a reference person, the Examiner’s findings do not show that Mansfield teaches or suggests the disputed limitations of recitations [4] and [6]. Appellant further argues that “the ‘algorithms’ of Yamada are not selected for each reference person.” Appeal Br. 12. The Examiner, in alternatively relying on Yamada to teach or suggest recitations [4] and [6], disagrees, finding that Yamada’s “matching algorithms are tie[d] to the quality of biometric data, where ‘biometric data of [a] reference person’ is broadly interpreted as ‘a registered fingerprint image of the registered user at the time of registration.’” Ans. 16 (some emphasis omitted) (citing Yamada ¶ 278). The Examiner similarly finds that Appeal 2020-005641 Application 14/437,803 7 “‘quality level’ of [the] biometric data of [a] reference person is broadly interpreted as ‘stor[age] therein in advance the quality for each registered fingerprint image.’” Id. at 17 (some emphasis omitted) (citing Yamada ¶ 279). The Examiner’s findings are not supported by evidence. Yamada merely determines “that some reduction of the quality is observed in the input fingerprint image recorded in the verification history table . . . from the quality of a registered fingerprint image of the registered user at the time of registration.” Yamada ¶ 278. Yamada alternatively makes a similar determination based on the ratios derived from “the number of the small areas determined as being the low-quality areas to the total number of small areas in . . . input fingerprint image[s].” Id. ¶ 279. When either the reduction in quality is observed or an average value or distribution based on the ratio determination exceeds a predetermined threshold, Yamada changes its verification process from a high-speed verification process to a precise verification process. Id. ¶¶ 278-80. But this selection of an algorithm applies to the biometric data being verified-the algorithm in Yamada is selected for the subject person, not for each reference person. Disputed recitation [4] is directed to dynamically selecting one of a first algorithm and a second algorithm “for each reference person” while recitation [7] is directed to, “for each reference person, comparing, by the computer, the biometric data of the type of the subject person with biometric data of the type of the reference person by using the selected one of the first algorithm and the second algorithm.” That is, an algorithm is selected for each reference person in the claimed method, and the algorithm for each reference person is used to compare biometric data of the subject person with biometric data for the Appeal 2020-005641 Application 14/437,803 8 reference person. The Examiner’s findings fail to show that Yamada’s selection of an algorithm for a subject person teaches or suggests selecting an algorithm for each reference person and applying the selection for each reference person in the manner of recitations [4] and [7]. Moreover, although the Examiner cites to Chan’s matching processes to teach recitation [7] (Final Act. 16-17 (citing, e.g., Chan ¶¶ 14, 26, 27, 40-42)), the Examiner’s findings do not show that Chan, even in combination with Yamada or Mansfield, teaches or suggests using an algorithm selected for a reference person in the manner of recitation [7]. In particular, Chan does not select between algorithms, but instead combines them to give a fused score. Chan ¶¶ 4, 14. The Examiner’s findings do not show that Robinson cures the noted deficiencies of Mansfield, Yamada, and Chan. Final Act. 28-29. Accordingly, we do not sustain the Examiner’s 35 U.S.C. § 103(a) rejection of claim 1, and the Examiner’s 35 U.S.C. § 103(a) rejections of claims 2-7, which contain similar recitations. CONCLUSION In summary: Claim(s) Rejected 35 U.S.C. § Reference(s)/Basis Affirmed Reversed 1-6 103(a) Chan, Mansfield, Yamada 1-6 7 103(a) Chan, Mansfield, Yamada, Robinson 7 Overall Outcome 1-7 REVERSED