PTAB

IPR2019-01029

General Electric Co v. Monument Peak Ventures LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Image Processing Method and Apparatus
  • Brief Description: The ’668 patent discloses automated methods and systems for processing digital images to emphasize the main subject. The technology aims to replace labor-intensive manual techniques by automatically identifying the main subject and altering the pixel values of the main subject and/or background to achieve a desired visual effect.

3. Grounds for Unpatentability

Ground 1: Obviousness over Kinjo-100 - Claims 1, 4, 6, 7, 9, 13-16, 23-25, and 27-29 are obvious over Kinjo-100

  • Prior Art Relied Upon: Kinjo-100 (Patent 5,978,100).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Kinjo-100, which describes a photographic printing apparatus, discloses every element of the challenged claims. Kinjo-100 teaches a computer-automated method to identify a "principal portion" of an image (the claimed "main subject") and modify the image based on that identification to improve its appearance for printing. The method involves segmenting an image into regions, assigning "weighting coefficients" to regions based on the probability of being the principal portion (the claimed "belief values" which form a "belief map"), and then calculating an "exposure amount" or "correction amount" to alter pixel density and color (the claimed "altering pixel values") to emphasize the identified principal portion. Petitioner asserted that this alteration follows the identification step, as required by the claims.
    • Motivation to Combine (for §103 grounds): As a single-reference ground, the argument was that a person of ordinary skill in the art (POSITA) would have found it obvious to implement the teachings of Kinjo-100, which inherently results in the claimed invention. For example, while Kinjo-100 describes an optional manual verification step, a POSITA would have understood that a fully-automatic process was desirable, particularly for processing large batches of images like a full roll of film.
    • Expectation of Success (for §103 grounds): A POSITA would have had a high expectation of success as Kinjo-100 provides a detailed blueprint for an automated system that performs all the key steps of the challenged claims.

Ground 2: Obviousness over Kinjo-100 in view of Etz - Claims 1, 4, 6, 7, 9, 13-16, 23-25, and 27-29 are obvious over Kinjo-100 in view of Etz

  • Prior Art Relied Upon: Kinjo-100 (Patent 5,978,100) and Etz (a 2000 publication titled "Ground Truth For Training and Evaluation of Automatic Main Subject Detection").
  • Core Argument for this Ground:
    • Prior Art Mapping: This ground was presented as an alternative, arguing that if Kinjo-100 alone was deemed insufficient, its teachings combined with Etz would render the claims obvious. Etz describes a detailed framework for automatically detecting the main subject of an image, which begins by segmenting pixels into regions and using probabilistic reasoning to generate a main subject belief map. Petitioner contended that Etz's teachings on image segmentation and belief value generation either supplement or make explicit what is already taught in Kinjo-100.
    • Motivation to Combine (for §103 grounds): A POSITA would combine Kinjo-100 and Etz because both address the same problem in the same technical field: automatic main subject detection in image processing. Petitioner argued that Kinjo-100 itself invites the use of other known "extracting methods" for identifying the principal portion, and Etz provides exactly such a known and applicable method. The combination would have been a predictable implementation of an improved main subject detection module within the broader image processing system of Kinjo-100.
    • Expectation of Success (for §103 grounds): A POSITA would have reasonably expected success in updating Kinjo-100's control unit to use the main subject detection algorithm from Etz, as it was a known technique designed for the same purpose, and its integration would yield the predictable result of improved subject identification.

4. Key Claim Construction Positions

  • "A computer method": Recited in the preamble of claims 1 and 14. Petitioner contended this term should be construed as "a method performed entirely by a computer (i.e., without manual intervention)." This construction was used to argue that the fully automated teachings of the prior art meet the claim limitations, distinguishing them from prior manual or semi-automated processes.
  • "main subject": Petitioner argued that the patent's specification provides three distinct definitions for this term: (1) a subjective definition based on an observer's opinion, (2) an objective definition based on a binary decision using a threshold on a belief map, and (3) a specific objective process using features like centrality and borderness. Petitioner asserted that the prior art renders the claims obvious under any of these three definitions.

5. Key Technical Contentions (Beyond Claim Construction)

  • Petitioner’s primary technical contention revolved around the multiple definitions of "main subject" in the patent’s specification. By demonstrating that the prior art met the claim limitations under all three potential interpretations (subjective, binary objective, and feature-based objective), Petitioner sought to preempt any narrow construction by the Patent Owner and show the claims were invalid regardless of the definition adopted. This strategy aimed to prove the robustness of the invalidity argument across different analytical frameworks.

6. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1, 4, 6, 7, 9, 13-16, 23-25, and 27-29 of the ’668 patent as unpatentable.