PTAB

IPR2026-00195

Amazon.com Inc v. InterDigital Inc

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Video Decoding and Encoding with Combined Resampling and Motion Compensation
  • Brief Description: The ’877 patent discloses methods and devices for video coding and decoding, addressing inefficiencies in adaptive streaming where a reference image may have a different size than the current image. The patent specification identifies that storing multiple, differently-sized versions of a reference image to handle resolution changes consumes scarce resources and risks overloading the decoded picture buffer. The purported invention combines the distinct mathematical operations of resampling and motion-compensation into a single, joint filtering process, thereby improving efficiency and reducing memory requirements by obviating the need to store resampled versions of reference images.

3. Grounds for Unpatentability

Ground 1: Claims 1, 4, 7, and 8 are obvious over Nakagawa in view of Sita.

  • Prior Art Relied Upon: Nakagawa (Patent 5,805,222) and Sita (Application # 2004/0150747).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Nakagawa disclosed a video coding system that adaptively switches between different picture resolutions (e.g., CIF and QCIF), thereby requiring the reconstruction of a current frame from a reference frame of a different size that is stored in a picture buffer. While Nakagawa established this problem, it did not specify the implementation details for the required resolution conversion during motion compensation. Sita allegedly supplied these missing details, teaching a decoder that handles resolution mismatches (e.g., high-definition vs. standard-definition) by using a combined resampling and motion-compensation operation. Petitioner contended that Sita’s disclosure of applying separate, one-dimensional horizontal and vertical filters that jointly perform both resampling and interpolation—as claimed by the equation GFv(s)=MCIFv(SCFv(s)), which represents a single filter composed of two functions—directly maps onto the core limitations of the challenged claims. Furthermore, Petitioner asserted both Nakagawa and Sita teach that no resampled version of the reference image is stored back into the picture buffer, meeting another key limitation.
    • Motivation to Combine: A POSITA would combine Sita's efficient, combined filtering technique with Nakagawa’s adaptive resolution system to provide a concrete implementation for the resampling process that Nakagawa required but did not detail. A secondary motivation was to upgrade Nakagawa's older H.261-based system to be compatible with more modern standards like MPEG-2, for which Sita's filtering operations were already designed, thus improving its utility.
    • Expectation of Success: Petitioner argued a POSITA would have a high expectation of success, as the combination involved implementing Sita's well-defined mathematical filtering solution to address a known need in Nakagawa's system, a task well within the ordinary skill in the art.

Ground 2: Claims 1, 4, 7, and 8 are obvious over Wiegand in view of Davies and Fimoff.

  • Prior Art Relied Upon: Wiegand (HEVC Working Draft 3, JCTVC-E603 v.8), Davies (JCTVC-F158 v.4), and Fimoff (Patent 6,628,714).

  • Core Argument for this Ground:

    • Prior Art Mapping: Petitioner asserted that Wiegand, a working draft of the HEVC standard, described a conventional video decoding process. Davies proposed modifying Wiegand to permit "resolution switching," allowing a decoder to predict frames across different resolutions by re-scaling reference pictures to adapt to fluctuating network conditions. This Wiegand-Davies combination established the context of decoding from a different-sized reference frame. However, this combination would perform resampling and motion compensation as separate steps, creating a known risk of overloading the limited memory of the picture buffer. Fimoff was presented as solving this exact problem by teaching a decoder that combines resampling and motion-compensation-interpolation into a single, joint filter operation specifically to reduce the amount of reference picture memory required when converting high-definition signals for standard-definition displays. Petitioner argued Fimoff’s teaching of a single filter performing both functions, applied successively on pixel columns and rows, and not storing resampled images in the buffer, fulfilled the key limitations of the challenged claims.
    • Motivation to Combine: A POSITA implementing the adaptive streaming system of Wiegand and Davies would confront the well-known problem of limited buffer memory. Petitioner contended this POSITA would have been motivated to incorporate the teachings of Fimoff, which provided an established solution to conserve memory by combining the separate resampling and motion compensation operations into a single, more efficient process for a predictable improvement.
    • Expectation of Success: Petitioner claimed a high expectation of success because Fimoff provided a clear description and equations to implement the combined filter. Integrating this known memory-saving technique into the Wiegand/Davies framework was a predictable design choice that involved combining known mathematical processes.

  • Additional Grounds: Petitioner asserted an additional obviousness challenge (Ground 3) combining Wiegand and Davies with Sita. This ground argued that Sita provided an alternative and equally compelling teaching to Fimoff for combining the resampling and motion-compensation operations into a single filter to address the buffer overload problem inherent in the Wiegand/Davies combination.

4. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1, 4, 7, and 8 of the ’877 patent as unpatentable.