PTAB

IPR2025-01521

Samsung Electronics America Inc v. SnapAid Ltd

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Real Time Assessment of Picture Quality
  • Brief Description: The ’325 patent describes a method and system for real-time assessment of picture quality in camera devices. The system uses multiple sensors to determine a plurality of quality indicators (QIs) which are then weighted and combined to calculate a total quality score for a captured image and provide suggestions to the user.

3. Grounds for Unpatentability

Ground 1: Obviousness over Anon, Takeuchi, Jasinski, Kosaka, Aisaka, and Garcia-Molina - Claims 1-4 and 7-10 are obvious over Anon in view of Takeuchi, Jasinski, Kosaka, Aisaka, and Garcia-Molina.

  • Prior Art Relied Upon: Anon (Patent 8,508,622), Takeuchi (Application # 2010/0149361), Jasinski (Application # 2012/0201427), Kosaka (Application # 2004/0012682), Aisaka (Application # 2010/0246939), and Garcia-Molina (a 2009 textbook).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Anon taught the foundational system of independent claim 1: an image capturing device with a processor that obtains a first quality value (QI1) from a motion/location sensor, estimates an associated weight, and computes an overall "composition measure" to provide real-time suggestions to the user. Petitioner contended the remaining limitations were disclosed by other references. Takeuchi was cited for teaching the calculation of a second quality value (QI2) based on image or face exposure. Jasinski was used to show the obviousness of dynamically estimating weights based on current image data, which Petitioner argued satisfied the "estimating" limitations. Kosaka was asserted to disclose obtaining a fourth quality value (QI4) responsive to lens obstruction, a common problem it was designed to detect. Aisaka was argued to teach analyzing an image via a deep learning algorithm to satisfy that limitation. Finally, Petitioner asserted that implementing Anon's "rule sets" in a pre-stored table, as taught by the Garcia-Molina textbook, was an obvious design choice.
    • Motivation to Combine: Petitioner contended a person of ordinary skill in the art (POSITA) would combine these references to improve Anon’s system. A POSITA would add Takeuchi's robust exposure analysis to create a more holistic quality assessment. Incorporating Jasinski’s dynamic weighting would allow Anon’s system to adapt to changing scene conditions. Adding Kosaka’s lens obstruction detection would address a common image flaw. Finally, using Aisaka’s deep learning methods for object recognition was a well-known technique for improving performance, and implementing suggestions in a table as shown in Garcia-Molina was a standard practice for data organization.
    • Expectation of Success: Petitioner argued a POSITA would have a reasonable expectation of success because the references described modular systems. Integrating known techniques like exposure analysis (Takeuchi), dynamic weighting (Jasinski), defect detection (Kosaka), and deep learning (Aisaka) into Anon's flexible framework was a straightforward extension that would yield predictable improvements in image quality assessment.

Ground 2: Obviousness over Anon, Takeuchi, Jasinski, Aisaka, and Cheatle - Claims 11, 14-17, and 20 are obvious over Anon in view of Takeuchi, Jasinski, Aisaka, and Cheatle.

  • Prior Art Relied Upon: Anon (Patent 8,508,622), Takeuchi (Application # 2010/0149361), Jasinski (Application # 2012/0201427), Aisaka (Application # 2010/0246939), and Cheatle (Application # 2002/0110286).

  • Core Argument for this Ground:

    • Prior Art Mapping: This ground challenged independent claim 11, which shared many limitations with claim 1. Petitioner again relied on Anon for the base system, Takeuchi for exposure and focus metrics, Jasinski for dynamic weighting, and Aisaka for using deep learning algorithms to analyze an image and obtain a quality value. To meet the limitation of obtaining a fourth quality value (QI4) based on recognizing a face as "known or unknown face, based on a pre-stored list of configured faces," Petitioner introduced Cheatle. Petitioner argued Cheatle explicitly disclosed a system that determines if a face is known by comparing it to a pre-stored database of facial features to assign a quality factor.
    • Motivation to Combine: Petitioner asserted a POSITA would be motivated to combine Cheatle with the Anon-based system to enhance user experience and create a more relevant image evaluation. By distinguishing known faces from unknown ones, the system could prioritize focus, exposure, or composition for recognized individuals, which was a logical and desirable improvement over a generic face detection system. This combination integrated complementary aspects of automated image quality assessment to create a more robust and comprehensive system.
    • Expectation of Success: Petitioner argued for a high expectation of success because both Anon and Cheatle described standard, compatible image processing techniques for facial detection and recognition. Integrating Cheatle's database-driven recognition into Anon's facial property analysis would involve conventional techniques and yield the predictable result of a more personalized and effective image quality assessment system.

  • Additional Grounds: Petitioner asserted additional obviousness challenges for the remaining dependent claims. These grounds relied on the same core combinations but added single references to meet specific limitations, such as adding Ramesh (Application # 2009/0296989) for estimating weights based on error analysis, Alhadef (Patent 8,009,198) for compensating for sensor drift, and Staudacher (Application # 2012/0105662) for combining face quality metrics based on a percentage of faces meeting a threshold.

4. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-20 of the ’325 patent as unpatentable under 35 U.S.C. §103.