PTAB

IPR2020-00757

Google LLC v. Uniloc 2017 LLC

Log In
Key Events
Petition
petition Intelligence

1. Case Identification

2. Patent Overview

  • Title: Method of Transcoding a Primary Encoded Signal
  • Brief Description: The ’960 patent discloses a method for transcoding a video signal from a higher bitrate to a lower bitrate. The invention improves upon prior art transcoders by adding a filtering step between the dequantizing and quantizing sub-steps to reduce noise and artifacts introduced during the transcoding process.

3. Grounds for Unpatentability

Ground 1: Obviousness over Keesman and Neri - Claim 1 is obvious over Keesman in view of Neri.

  • Prior Art Relied Upon: Keesman (a 1996 journal article on transcoding MPEG bitstreams) and Neri (a 1994 journal article on filtering in the DCT domain).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Keesman disclosed a standard video transcoder architecture for reducing bitrate, which included decoding steps (dequantizing) followed by encoding steps (quantizing). However, Keesman did not explicitly disclose a filtering step between its dequantizer (DQ1) and quantizer (Q2). Petitioner asserted that Neri remedied this deficiency by teaching an "inter-block" filter that operates directly on coefficients in the Discrete Cosine Transform (DCT) domain to reduce artifacts during video format conversion. The signal in Keesman between the dequantizer and quantizer is in the DCT domain, making it an ideal location to apply Neri’s filter.
    • Motivation to Combine: A person of ordinary skill in the art (POSITA) would combine Neri’s filter with Keesman’s transcoder to improve the perceived image quality of the lower-bit-rate output video, a well-known objective in the field. Neri explicitly recommended using a low-pass filter to reduce aliasing effects when converting video formats. Petitioner contended that inserting Neri's filter into Keesman's architecture was a straightforward application of a known technique to solve a known problem. The petition further argued that the resulting filter would be "recursive" as claimed, because Keesman's architecture uses a motion-compensation feedback loop where previously filtered data is used in future calculations.
    • Expectation of Success: A POSITA would have had a reasonable expectation of success because combining the references involved a limited number of predictable design choices. Placing Neri's DCT-domain filter at a point in Keesman's system where the signal is already in the DCT domain would predictably reduce distortion, achieving the intended result.

Ground 2: Obviousness over Keesman and Kim - Claims 4 and 5 are obvious over Keesman in view of Kim.

  • Prior Art Relied Upon: Keesman (a 1996 journal article) and Kim (Patent 6,249,549).

  • Core Argument for this Ground:

    • Prior Art Mapping: This ground challenged claims 4 and 5, which require the filtering step to be a "spatial filtering step" that is "only applied to intra-coded macroblocks." As in Ground 1, Keesman provided the foundational transcoder. Kim was cited for its disclosure of an "intra-block lowpass filter" that operates in the frequency (DCT) domain to reduce aliasing distortion. Petitioner argued that Kim's filter constitutes a "spatial filter" in the context of the ’960 patent (i.e., non-temporal). Critically, Kim specified that its filter operation "is performed on an intra-block basis," directly teaching the limitation of applying the filter only to intra-coded macroblocks. For claim 5, which adds detecting a macroblock's type and applying the filter based on a corresponding "label," Petitioner argued Keesman's inherent switching functionality for different macroblock types (I, P, or B frames) met this limitation when combined with Kim.
    • Motivation to Combine: A POSITA would have been motivated to integrate Kim’s intra-block filter into Keesman’s transcoder for the same general purpose of improving image quality. Kim taught that filtering in the DCT domain was computationally efficient and effective at reducing distortion. Applying the filter only to intra-coded blocks was a known technique for targeted image improvement without affecting motion-compensated blocks.
    • Expectation of Success: A POSITA would have expected this combination to work predictably. Implementing Kim's filter in Keesman’s system was a known method to improve a similar device, yielding the predictable result of reduced distortion.

  • Additional Grounds: Petitioner asserted additional obviousness challenges, including that claim 1 is obvious over Keesman, Neri, and Dubois (a 1984 journal article) to further support the specific recursive filter equation recited in the claim. Petitioner also argued that claims 4 and 5 are obvious over Keesman, Kim, and Matsumura (Patent 6,792,045) to further teach a control system for adaptively selecting a filter based on macroblock type.

4. Key Claim Construction Positions

  • "transformed signal[s]" / "transformed coefficients": Petitioner proposed these terms mean "data concerning video that has been discrete cosine transformed and inverse quantized." This construction was important to show that the signals in the prior art at the point of filtering (e.g., after Keesman's dequantizer) matched the claimed signal type.
  • "intended to use a recursive filter such as...": Petitioner argued that the claim language "intended to use" and "such as" is exemplary and does not limit the claim to the specific filter equation that follows. This position was critical for the primary argument in Ground 1, which established the recursiveness of the combined system through its feedback loop rather than by explicitly mapping every term of the patent's example equation to the prior art.

5. Relief Requested

  • Petitioner requested institution of an inter partes review (IPR) and cancellation of claims 1, 4, and 5 of Patent 7,012,960 as unpatentable.