PTAB

IPR2020-00044

Netflix Inc v. Uniloc 2017 LLC

Key Events
Petition

1. Case Identification

2. Patent Overview

  • Title: Macroblock-Based Object-Oriented Video Coding
  • Brief Description: The ’229 patent describes a method for object-oriented video compression. The method involves dividing a video image into a stationary background region and a foreground object region on a macroblock-by-macroblock basis, rather than a pixel-by-pixel basis, to encode each region separately and more efficiently for low bit-rate applications.

3. Grounds for Unpatentability

Ground 1: Obviousness over Talluri in view of Lin - Claims 1, 3, 5, and 6 are obvious over Talluri in view of Lin.

  • Prior Art Relied Upon: Talluri (a 1997 publication titled A Robust, Scalable, Object-Based Video Compression Technique for Very Low Bit-rate Coding) and Lin (Patent 6,192,148).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Talluri teaches a complete object-based video compression system that aligns with the core steps of claim 1. Talluri disclosed segmenting a video into a "stationary background" and "moving objects" on a macroblock basis using image differencing between frames (meeting claim 1(a)). It further taught using bitmap masks to represent the shape of object regions (meeting claim 1(b)) and encoding the object region using motion compensation and a DCT-based approach (meeting claim 1(c)). While Talluri disclosed encoding the background, Lin was introduced to explicitly teach the "wherein" clause of claim 1. Lin described a method for determining when to skip macroblocks in MPEG-2 encoding, specifically for static image portions, by comparing pixel differences between co-located blocks in current and previous frames against a threshold. If the difference was below the threshold, the block was skipped, and the previous frame's block was reused, directly teaching the reuse of corresponding macroblocks for stationary regions as claimed.
    • Motivation to Combine: A Person of Ordinary Skill in the Art (POSA) would combine Talluri with Lin to improve the efficiency of Talluri's system. Talluri’s goal was efficient compression for low bit-rate applications like videoconferencing, which often feature static backgrounds. Lin provided a well-known and advantageous technique to efficiently handle such static regions by skipping macroblocks, thereby reducing computation and bit rate. A POSA would have found it obvious to apply Lin’s specific skipping method to the "stationary background" regions explicitly identified by Talluri to achieve greater coding efficiency.
    • Expectation of Success: As both references described techniques for standard-compliant, macroblock-based video encoders (MPEG-4 and MPEG-2, respectively), a POSA would have had a high expectation of success in combining Lin's block-skipping feature into Talluri's hybrid coder, as it was a predictable integration of known, compatible technologies.

Ground 2: Obviousness over Monro in view of Chai - Claims 1, 5, and 6 are obvious over Monro in view of Chai.

  • Prior Art Relied Upon: Monro (WO 98/11730) and Chai (a 1997 publication titled Foreground/Background Video Coding Scheme).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner asserted that Monro disclosed an object-oriented video system that separated a foreground object from a background object using block-based segmentation and a difference comparison (mean squared error) against an "accepted background" from previous frames. If a current block matched the accepted background within a threshold, it was tagged as "known background" and reused by the decoder without being re-transmitted, teaching the core background reuse concept of claim 1. However, Monro described segmentation using smaller (e.g., 4x4) blocks. Chai was introduced to provide the motivation to adapt Monro's system to the macroblock level. Chai taught a segmentation scheme for H.263-based encoders and explicitly stated that segmentation resolution should be reduced to the macroblock level to ensure compatibility with the encoder.
    • Motivation to Combine: A POSA would combine Monro with Chai to integrate Monro's novel object-separation logic into a standard-based encoder. Monro allowed for "any convenient coding" scheme, and Chai provided a clear roadmap for integrating a similar foreground/background segmentation approach with a standard H.263 encoder, which was a logical choice for the low bit-rate applications targeted by both references. Chai’s teaching to perform segmentation at the macroblock level for compatibility would make it obvious to modify Monro’s block-based partitioning to a macroblock-based one, thus rendering the method of claim 1.
    • Expectation of Success: Combining Monro's segmentation logic with a standard-based H.263 encoder as guided by Chai would be a straightforward engineering task. The benefits of leveraging a well-defined standard like H.263 would have provided a strong expectation that the resulting system would be robust and effective.

4. Key Claim Construction Positions

  • "generating coded pixel information of a previous frame macroblock..." (Claim 1, step d): Petitioner argued this phrase is ambiguous. It could mean either (1) encoding blocks of the previous frame that correspond to the background of the current frame, or (2) encoding background blocks of the current frame using coded data from the previous frame. Petitioner contended that the prior art disclosed both interpretations, making the limitation obvious regardless of the construction adopted.
  • "coded data coded shape information..." (Claim 1, step e): Petitioner asserted that this phrase contains a typo. A POSA would understand the phrase "coded data" to be a header for the list that follows, meaning the claim should be read as "coded data including coded shape information, coded object pixel information, and coded stationary pixel information." This interpretation is supported by the patent's specification.

5. Relief Requested

  • Petitioner requested institution of an inter partes review (IPR) and cancellation of claims 1, 3, 5, and 6 of the ’229 patent as unpatentable.