GEVC patent in SISVEL AV1 pool appears not essential

As part of an ongoing series examining the patent holders and pools erroneously designating patents as essential, we highlight U.S. Patent 10,460,344 titled “Region merging and coding parameter reuse via merging.” This patent is owned by GE Video Compression (GEVC). GEVC has designated the ’344 patent as essential to the AV1 standard as a part of SISVEL’s AV1 Patent Pool. See AV1 Patent List, AV1 Family AV1-040, available at https://www.sisvel.com/images/documents/Video-Coding-Platform/PatentList_AV1.pdf.

GEVC’s U.S. Patent 10,460,344 should not be considered to be essential to the AV1 standard. The ’344 patent is directed to a decoder that uses a merging or grouping of simply connected regions using a reduced amount of data. ’344 patent, Abstract. Namely, a merge indicator indicates whether a region currently being decoded should be reconstructed based on a motion coding parameter. If the indicator indicates copying, the appropriate vector is copied. If the indicator indicates compute, the appropriate motion vector is computed.  Id., claims 1, 9, 17, 26. 

The concept of a merge indicator is an evolved form of motion vector competition. See, e.g., Joel Jung and Guillaume Laroche, “Competition-Based Scheme for Motion Vector Selection and Coding,” VCEG Contribution VCEG-AC06r1, Klagenfurt, Austria, July 2006. In contrast to the ’344 patent and prior motion vector competition literature, the AV1 standard does not employ a merge indicator; rather, the concepts of merging and computing a motion vector is spread over multiple values, not merely an indicator to either copy the ap or compute the motion vector.  See, e.g., AV1 §§ 5.11.26 (assign_mv syntax code used to limit the maximum size of motion vectors); 5.11.23 (syntax); 6.10.22 (semantics describing new_mv, zero_mv, and ref_mv);. 

Thus, the ’344 patent does not appear to be essential to the AV1 standard despite being declared as essential. The public would benefit from appropriate scrutiny of patent pools that allegedly cover critical technical standards, particularly open-source standards such as AV1.

Another InterDigital patent in SISVEL AV1 pool appears not essential

As part of an ongoing series examining the patent holders and pools erroneously designating patents as essential, we highlight U.S. Patent 9,674,556 titled “Methods and apparatus for in-loop de-artifact filtering.” This patent is owned by InterDigital VC Holdings, Inc. InterDigital has designated the ’556 patent as essential to the AV1 standard as a part of SISVEL’s AV1 Patent Pool. See AV1 Patent List, AV1 Family AV1-056, available at https://www.sisvel.com/images/documents/Video-Coding-Platform/PatentList_AV1.pdf.

InterDigital’s ’556 patent should not be considered to be essential to the AV1 standard. The ’556 patent is directed to in-loop artifact filtering, and in particular, two filters for successively performing in-loop filtering. The claims of the ’556 patent require (1) “a deblocking filter for performing a first pass to reduce blocking artifacts” and (2) “an adaptive sparse de-noising filter for performing a second pass to reduce noise.” ’556 patent, claim 1.

The AV1 standard uses a de-ringing filter called CDEF as its second filter, not the ’556 patent’s claimed “de-noising filter.” In detail, the AV1 standard employs a distinct process known as the CDEF (constrained directional enhancement filter) process. According to the AV1 standard, the “purpose of the CDEF is to perform deringing….” See AV1, § 7.15 (emphasis added). Contemporaneous research confirms the CDEF works to reduce ringing artifacts. See Midtskogen et al., “The AV1 Constrained Directional Enhancement Filter (CDEF),” arXiv:1602.05975 [cs.MM], Feb. 18, 2016, p. 1, available at https://doi.org/10.48550/arXiv.1602.05975.

Although InterDigital cites the CDEF provisions of the AV1 standard as part of its claims for essentiality, the applicant expressly disclaimed de-ringing filters from the scope of its claims. The ’556 Patent explicitly distinguishes between de-ringing filters (e.g., CDEF) and its claimed “de-noising” filters, listing them as separate filtering options: “At least one of the filters includes, for example, a deblocking filter, a de-ringing filter, a de-noising filter…” ’556 patent, 10:13–16. The difference was confirmed by the applicant during prosecution of the ’556 patent’s parent application, U.S. Patent Application No. 12/312,386, in which the applicant stated that an applied reference’s “deringing filtering does not disclose or suggest a sparse de-noising filter for performing a second pass to reduce noise,” which is similarly included in the ’556 Patent’s claims. See Prosecution History of U.S. Patent Application 12/312,386, Response to Office Action filed June 3, 2014, p. 10 (italics in original).

Moreover, AV1 standard’s use of (1) a deblocking filter followed by (2) de-ringing filter was already well known in the art prior to the ’556 Patent. For example, U.S. Patent 7,738,563 to Pelc, filed over two years prior to the ’556 Patent’s earliest priority date, disclosed a “filter 430 [that] applied [1] deblocking filtering operations and then [2] de-ringing filtering operations.” Pelc ‘563, 5:49-52. Similarly, U.S. Patent 8,537,903 to Lim, filed over a year prior to the ’556 Patent’s earliest priority date, disclosed a decoding system with “[1] a de-blocking system configured to receive decoded video data and decoding information and remove blocking artifacts from the decoded video data, and [2] a de-ringing system configured to remove ringing artifacts from the de-blocked video data.” Lim ‘903, Abstract. Again, this is similar to the operations described in the AV1 standard and distinct from a second-stage “adaptive sparse de-noising filter” in the ’556 Patent.

Thus, the ’556 Patent does not appear to be essential to the AV1 standard even though it has been declared essential and actively licensed as being so. The public would benefit from appropriate scrutiny of patent pools that allegedly cover critical technical standards.

The ’556 Patent is related to U.S. Patent 9,277,243, which InterDigital has also declared essential, but whose non-essentiality is reviewed in this post, https://www.unifiedpatents.com/insights/2023/6/15/interdigital-sisvels-av1-pool-us-9227243.

InterDigital patent in SISVEL AV1 pool appears not essential

As part of an ongoing series examining the patent holders and pools erroneously designating patents as essential, we highlight U.S. Patent 9,277,243 titled “Methods and apparatus for in-loop de-artifact filtering.” This patent is owned by InterDigital VC Holdings, Inc. InterDigital has designated the ’243 patent as essential to the AV1 standard as a part of SISVEL’s AV1 Patent Pool. See AV1 Patent List, AV1 Family AV1-056, available at https://www.sisvel.com/images/documents/Video-Coding-Platform/PatentList_AV1.pdf.

InterDigital’s ’243 Patent should not be considered to be essential to the AV1 standard. The ’243 Patent contains a single claim directed to in-loop artifact filtering, and in particular, two filters for successively performing in-loop filtering:  (1) “a deblocking filter for performing a first pass to reduce blocking artifacts” and (2) “an adaptive sparse de-noising filter for performing a second pass to reduce noise.” ’243 Patent, claim 1.

The AV1 standard uses a de-ringing filter called CDEF as its second filter, not the ’243 Patent’s claimed “de-noising filter.” In detail, the AV1 standard employs a distinct process known as the CDEF (constrained directional enhancement filter) process. According to the AV1 standard, the “purpose of the CDEF is to perform deringing….” See AV1, § 7.15 (emphasis added). Contemporaneous research confirms the CDEF works to reduce ringing artifacts. See Midtskogen et al., “The AV1 Constrained Directional Enhancement Filter (CDEF),” arXiv:1602.05975 [cs.MM], Feb. 18, 2016, p. 1, available at https://doi.org/10.48550/arXiv.1602.05975.

Although InterDigital cites the CDEF provisions of the AV1 standard as part of its claims for essentiality, the applicant expressly disclaimed de-ringing filters from the scope of its claims. The ’243 Patent explicitly distinguishes between de-ringing filters (e.g., CDEF) and its claimed “de-noising” filters, listing them as separate filtering options: “At least one of the filters includes, for example, a deblocking filter, a de-ringing filter, a de-noising filter…” ’243 Patent, 10:2-5. The difference was confirmed by the applicant during prosecution of the ’243 Patent’s application, U.S. Patent Application No. 12/312,386, in which the applicant stated that an applied reference’s “deringing filtering does not disclose or suggest a sparse de-noising filter for performing a second pass to reduce noise,” which is similarly included in the ’243 Patent’s claims. See Prosecution History of U.S. Patent Application 12/312,386, Response to Office Action filed June 3, 2014, p. 10 (italics in original).

Moreover, AV1 standard’s use of (1) a deblocking filter followed by (2) de-ringing filter was already well known in the art prior to the ’243 Patent. For example, U.S. Patent 7,738,563 to Pelc, filed over two years prior to the ’243 Patent’s earliest priority date, disclosed a “filter 430 [that] applied [1] deblocking filtering operations and then [2] de-ringing filtering operations.” Pelc ‘563, 5:49-52. Similarly, U.S. Patent 8,537,903 to Lim, filed over a year prior to the ’243 Patent’s earliest priority date, disclosed a decoding system with “[1] a de-blocking system configured to receive decoded video data and decoding information and remove blocking artifacts from the decoded video data, and [2] a de-ringing system configured to remove ringing artifacts from the de-blocked video data.” Lim ‘903, Abstract. Again, this is similar to the operations described in the AV1 standard and distinct from a second-stage “adaptive sparse de-noising filter” in the ’243 Patent.

Thus, the ’243 Patent does not appear to be essential to the AV1 standard even though it has been declared essential and actively licensed as being so. The public would benefit from appropriate scrutiny of patent pools that allegedly cover critical technical standards.

The Adoption and Benefits of Royalty-Free Licensing

As part of the efforts of Unified Edge, John Jarosz and his team of economists at the Analysis Group have published a paper on the benefits of royalty-free licensing. Read the abstract below and follow the link to download the paper.

Abstract

Much licensing of patented technology is royalty-bearing, allowing innovators to be compensated for their investments in researching, developing, and then in commercializing the resulting products.

However, over the last couple of decades, standards-based technologies have expanded, in part, due to the increased need for device interoperability, communication, and replacement. Though some of the licensing of patents covering such technologies bears royalties, royalty-free licensing of standards-based technology has grown increasingly common. Indeed, perhaps only a small percentage of today’s widely adopted standards-based technologies are subject to royalty payments. Furthermore, there has been a proliferation of non-standards-based patented technologies and protocols that companies offer on a royalty-free basis (e.g., Adobe PDF and Thunderbolt, discussed further below).

The advantages of royalty-free licensing of technology are obvious for competition broadly and can be numerous for both consumers and companies, specifically, those who benefit from, inter alia, reduced costs and lowered barriers to market entry. But royalty-free licensing can be appealing to innovators as well, as it can result in faster adoption, increased usage, and further innovations that incorporate the standardized technologies. Using some well-known examples, this paper examines the widespread adoption of, motivations behind, and benefits associated with royalty-free licensing in standards-based settings, and even, in some cases, outside of a formal standard-setting organization (“SSO”). Though royalty-free licensing has many perceived and actual advantages, it is not the optimal route for all technology licensing. It can be and is, of course, optimal in many settings.

Find the complete report HERE.

Open source rules engine for patent prosecution, PATCEPTA, kick off on Feb 9

Unified Patents and the Linux Foundation Announce Patcepta

Original article from THE LINUX FOUNDATION | 30 NOVEMBER 2022

Patcepta is a new open source rules engine for improving patent prosecution and management through automation.  Any interested party can join to help foster a new generation of IP management practices using open source software.

Unified Patents decided to work with The Linux Foundation in making the Patcepta project open source to remove critical barriers to entry that attorneys and entrepreneurs face in managing their patent portfolio. This project is the first of its kind in that it provides an open, understandable, and auditable ruleset and toolkit for enabling more efficient and innovative processes and tools for managing IP portfolios.

“We are excited to see this fundamental capability become an open source project for collaboration and shared dependency'' said Mike Dolan, Senior Vice President of Projects at the Linux Foundation. “Jurisdiction rules and a common rules engine are shared challenges faced by in-house and outside counsel trying to manage patent portfolios - a perfect opportunity to develop collaboratively and maintain in an open source community.”

“A key barrier to entry was made very clear in the beginning around docketing rules, which just in the US are over 500, while only about 80 represent almost 80% of the ones used,” said Shawn Ambwani, Co-Founder and COO of Unified Patents. “The number of rules is one of the reasons why the market is so fragmented and why the software is mainly provided as a loss leader to lock in services contracts for paying patent annuities. Another issue IP managers face is concerns associated with bad rules or missing a deadline – now the rules data and code will be transparent and open for any IP manager to see or update.  Finally, integrating such docketing software into larger processes and accounting systems makes switching costs high.  

Software providers, law firms, and even some corporations develop their code and maintain it as the rules change, which can be expensive and repetitive. 

After talking to The Linux Foundation (TLF) and a few members to gauge interest, we decided a potential course would be to help make the rules / code open source so anyone can build on top of it. In addition, having open source allows for collaboration and updates and, most importantly, allows everyone to see the code to get confidence that it is safe to rely upon.

Ultimately, the goal is to foster better and more innovative patent management processes by eliminating a major barrier.  

Patcepta is now an official project hosted by the Linux Foundation. You can view the project here or go to www.patcepta.org. It is the first legal industry codebase, open rules data, and project ever hosted by the foundation. We hope it will not only complete publishing all the United States rules and code, but there are opportunities for the open source community to add international jurisdictions and perhaps an interface down the road.”

We hope all parties interested in making patent and IP management easier evaluate and contribute to this project and help remove docketing rules as a barrier to managing their portfolio.

An initial kickoff meeting is scheduled for February 9th, 10a PT, so participants can join and help in moving this project forward or getting access to the repository. Please contact info@patcepta.org for any questions.