`571-272-7822
`
`
`
`
`Paper No. 40
`Date: August 18, 2020
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`UNIFIED PATENTS, LLC,
`Petitioner,
`
`v.
`
`VELOS MEDIA, LLC,
`Patent Owner.
`____________
`
`IPR2019-00757
`Patent 9,930,365 B2
`____________
`
`Record of Oral Hearing
`Held: June 16, 2020
`____________
`
`
`
`
`Before MONICA S. ULLAGADDI, JASON W. MELVIN, and
`AARON W. MOORE, Administrative Patent Judges.
`
`
`
`
`
`
`
`
`
`
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`
`
`APPEARANCES:
`
`ON BEHALF OF THE PETITIONER:
`
`
`ERIC A. BURESH, ESQ.
`7015 College Blvd.
`Suite 700
`Overland Park, KS 66211
`
`
`
`ON BEHALF OF THE PATENT OWNER:
`
`
`BARRY J. BUMGARDNER, ESQ.
`CHRIS BRANNIGAN, ESQ.
`KENNY MCCLURE, ESQ.
`Nelson Bumgardner Albritton, P.C.
`3131 West 7th St., Suite 300
`Fort Worth, TX 76107
`
`
`
`
`The above-entitled matter came on for hearing on Tuesday, June 16,
`
`2020, commencing at 9:00 a.m. EDT, by video/telephone.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`2
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`
`P R O C E E D I N G S
`- - - - -
`JUDGE ULLAGADDI: Good morning and welcome to the Patent
`Trial Appeal Board. We are here today for oral arguments and interparty
`reviews matter number 2019-00757. A case in which Unified Patents is the
`Petitioner and Velos Media is the Patent Owner. At issue is U.S. Patent
`number 9,930,365.
`Your panel for the hearing today includes myself, Judge Ullagaddi,
`Judge Melvin and Judge Moore. I would like to start by getting the
`appearances of counsel. Who do we have on behalf of Petitioner?
`MR. BURESH: Your Honor, this is Eric Buresh of Erise IP on behalf
`of Unified Patent, LLC, and with me and also who will be appearing is
`David Cavanaugh of WilmerHale on behalf of Unified Patent.
`JUDGE ULLAGADDI: Thank you, Mr. Buresh. Who do we have on
`behalf of Patent Owner?
`MR. BUMGARDNER: Good morning, Your Honor. This is Barry
`Bumgardner here on behalf of Patent Owner with me here physically and
`present in my office is Chris Brannigan, another attorney of record who will
`be discussing some real party interest issues. Also, present with me in the
`office is Kenny McClure in-house counsel with Velos and then also on the
`dial-in line is another in-house attorney with Velos: Rory Litten.
`JUDGE ULLAGADDI: Thank you, Mr. Bumgardner. And thank
`you all for joining us today. I've got a few administrative details about the
`procedure and format of the hearing that I would like to go over before we
`get started. When you are working through your demonstrative slides,
`please make sure that you reference the demonstrative slides that you are
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`3
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`using because we can't see them.
`Each party will have 60 minutes total to argue their case. As Patent
`Owner has requested a bifurcated trial, we are going to start with the
`nonconfidential portion of trial. We'll first hear from Petitioner. Petitioner,
`you will present your arguments related to patentability.
`Patent Owner, you will then be permitted to present your argument.
`Petitioner would you like to reserve any time for rebuttal today?
`MR. BURESH: As to the ethical side of the nonconfidential side,
`Your Honor, I would like to reserve 15 minutes.
`JUDGE ULLAGADDI: Fifteen minutes reserved on rebuttal; okay.
`MR. BURESH: And just to clarify, Your Honor, I intend to go 35
`minutes in my opening and approximately 15 minutes in rebuttal and then
`we'll reserve approximately 10 minutes for our RPI rebuttal.
`JUDGE ULLAGADDI: Thank you. When you are ready you may
`begin.
`MR. BURESH: Thank you, Your Honor. I am going to be working
`with Petitioner's demonstrative slides which you should have in front of you,
`starting at slide number 2 just marked B. I'm going to be discussing first
`partitioning encoding and then I'm going to provide a little background of
`the 365 patent, the challenged patent.
`Then I will discuss how partitioning occurs in the primary prior art
`reference -- and we'll address the key prime limitations in that portion - there
`will be a short portion on Novotny's A and B syntax element called for by
`the challenged claims - and then I will be wrapping up.
`Moving to slide number 3 in reviewing coding, Your Honors, the
`purpose of partitioning is really to simplify it down as much as I can, bigger
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`4
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`blocks are better. They're more efficient, but when video images or picture
`images have granule area and particularly are in demand and particular
`features that will be changing from image to image, it becomes necessary to
`utilize smaller blocks in order to capture the fine nature of those edge pieces.
`So, on slide number three, we have an example of what might be an
`ideal partitioning where you see background information is depicted and
`captured through larger blocks and then when you get to edge features where
`there will be more change from image to image, you go or work your way
`down to smaller blocks, which would be partitioned off the larger blocks.
`This is depicted in figure 12 of the 365 patent where you see the idea
`of larger blocks end-to-end and working or segmenting down to smaller
`blocks within that larger block. Also, depicted on figure 12 is the types of
`decoding modes that might be used for the respective blocks and smaller
`blocks within those blocks.
`That is the underlying purpose for partitioning. At this timeframe,
`Your Honors, virtually all decoding and coding is image including the main
`encoder and decoder we utilize - what is called rate distortion analysis - to
`determine the proper sizing of what is desirable sizing of blocks and what
`rate distortion analysis is effectively a cost benefit where you learn how
`much quality of video are you losing in order to go with a larger block that is
`marked “essential patent.”
`So, you're looking at both the quality size and the efficiency size and
`using an analysis. There are a number of different analyses that can be used
`by using two-piece analysis to find the best sizing for your blocks. And if
`we go to slide number 6 after slide number 5 outline, you will see in the 365
`patent and before I go any further on slide number 6, many of the Patent
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`5
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`Owner arguments in this case are what I would call “higher arguments”
`where the patent owner is attempting to distinguish in a broad brush fashion
`between prior referenced Kalker and the 365 patent.
`And one of the impressions, at least I received, when the patent owner
`responded and also the patent owner served, was a suggestion that the 365,
`and this is drawn from the fact that the claims of the 365 in this particular
`application are all directed to the decoder. But the suggestion was that the
`365 has potentially very intelligent decoders and was able to make
`segmentation additions on its own, separate and apart from the encoders.
`And one of the purposes of these next few slides is just to debunk that.
`The 365 patent is the same basic structure as all systems in the timeframe
`where the encoder is making the decision. The encoder is intact information
`to the decoder and the decoder is simply reversing the process of the
`encoder. And that reversing of the decisions that were made by the encoder
`is what the 365 patent refers to as partitioning.
`So, the partitioning decision is made on the encoder side. Syntax
`information is transmitted to the decoder and the decoder reverses using that
`syntax information to recreate an image and that uses data in itself
`partitioning at the decoder. And we'll see that -
`JUDGE ULLAGADDI: May I ask a question about that?
`MR. BURESH: Sure.
`JUDGE ULLAGADDI: If the encoder side is doing the partitioning
`then why would the decoder side also partition? Wouldn't it be putting
`everything back together?
`MR. BURESH: Yeah, and that's the point, Your Honor. The decoder
`side in the act of - as the words you use "putting everything back together" it
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`6
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`is recreating a segmentation after a partition map and that act of taking the
`video image that in its original data stream is that the whole image and
`bringing it in through the decoder and breaking it up into blocks that will be
`processed on a block -
`That's efficient of the image blocks is the partitioning on the decoder
`side. Now, as I said, the decoder is not making the decision about what the
`block size will be, but when the decoder devised the image into blocks the
`process then is exactly the same way and in exactly the same block sizes as
`the encoder is partitioning. And that is what is described in the 365 patent.
`JUDGE ULLAGADDI: Thank you.
`MR. BURESH: And if we go to slide 6, the main point here - a
`couple of points, one the encoder in the 365 patent is determining the
`different partitioning levels for the macro block. Just like in Kalker it is the
`encoder that is making the determination or the partitioning in the 365 patent
`and that determination that the partition levels as well as decoding modes to
`apply to the partition is based on a rate distortion analysis.
`I highlight that just to show the similarities between the 365 and
`Kalker which we'll see in a minute, uses exactly that same rate distortion
`analysis to make the segmentation decision at the encoder. Turning to slide
`number seven, I'm going to go quickly through these next two.
`Again, just to show a correlation between the 365 patent and Kalker,
`the 365 patent on slide number 7 is talking about the encoder selecting the
`syntax or excuse me block base syntax to use based on the largest block.
`We will see also the Kalker's that once it's blocked at least in the
`embodiment we are relying upon in the partition is also using the largest
`block syntax.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`7
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`
`And again, the video decoder turning to slide number 8, is receiving
`that syntax information from the video encoder and is similarly using it to
`select the same block-based syntax information based upon the largest
`blocks. Again, we see between slides 7 and 8 that the decision made at the
`encoder to the decoder to recreate.
`And I'm going to get to the little more granular on that as we go
`because I think it's important that it's not just in the 365 patent turning to
`slide nine, it is not just a coded unit syntax information. There is syntax
`information in the 365 that is sent from the encoder to the decoder that
`actually shows the partitioning of individual blocks as well.
`So, you have the largest block size and smallest block size that is
`transmitted in the 365 patent, but you also have partitioning for the
`individual blocks. Turning to slide 10, we're going to see what the syntax
`information is and what it reflects from slide 9. In slide 10, again, you're
`looking at the video encoder side and what it is generating in a second block
`on slide 10.
`The video encoder is generating what type syntax information that
`indicates the size of a block and identifies the partitioning in the encoding
`mode used to encode the partition. So, we see again, the encoder in the 365
`patent is making the partitioning decision. It is generating syntax
`information to reflect those partitioning decisions and it is transmitting those
`partitioning syntax - that particular syntax information to the decoder so that
`the decoder can itself reverse the process which the 365 calls partitioning.
`And while the 365 does not go into any great detail about the
`decoders, the vast majority of the description in the 365 relates to the
`encoder, but if we turn to slide 11 and look at the pending claim three which
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`8
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`depends on challenge claim one we see that the decoder receives and further
`comprising decoding one or more syntax elements representative of
`partitioning of the current block. So, in claim one we have the concept of
`the first syntax element that is the smallest block in the encoded unit, a
`second syntax element that is the largest block in the encoded unit and the
`point here is there is additional syntax information that is actually
`representing the partitioning for the current block. So, between the largest
`and the smallest how a particular block is partitioned is based upon
`information sent from the decoder - encoder and received by the decoder.
`Any questions on that, Your Honors?
`JUDGE ULLAGADDI: No.
`MR. BURESH: One more point before we leave the 365 patent on
`slide number 12, this is responsive to the arguments by Patent Owner that
`Kalker uses the words "blocks and smaller blocks" - use the word
`“subblocks” in Kalker and the point of slide 12 is simply to note that within
`the 365 patent the notion of a “subblock” is also a block. You will always
`have blocks. You will have larger blocks you can partition into smaller
`blocks.
`The smaller blocks can be called subblocks, but those subblocks can
`just be called smaller blocks. The nomenclature there, there is nothing
`special about subblocks except that they are a smaller block within a larger
`block.
`Skipping over slide 13 which simply indicates that I'm moving onto
`Kalker now. Slide number 14 the disclosing in Kalker again, the basics if
`you just look at figure one, it's a very easy kind of reference here in Kalker
`to get the basic framework where you have a second circuit that generates a
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`9
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`second of video images into blocks.
`You have a map encoder that generates your segmentation map based
`upon the segmentation decided by the segmentation circuit. That
`information is transmitted to a decoder which again reverses the process and
`gives you a segmentation map that can be applied to inverse the operation on
`the encoder side.
`JUDGE ULLAGADDI: I think I'd like to come back to that point. If
`the decoder side is inversing as you say operations made on the encoder
`side, what is the inverse of partitioning?
`MR. BURESH: So, if you look at figure number1 in Kalker which is
`tighter on the page in the figures, but in figure number 1 you see the concept
`of the -- coming out of the segmentation circuit. That is -- what the
`segmentation circuit does is it takes an input whole video image and it
`determines which is on slide number 14 actually based upon that rate
`distortion analysis the same one in the 365 patent, it determines at the
`segmentation circuit the optimal block size.
`That optimal block size is then encoded by the map encoders. So, you
`have your block sizes that are desirable, the map encoder will generate
`syntax information to reflect that in a segmentation map. When the map
`decoder receives that information that encoded segmentation map what it
`does is takes -- it generates a new segment or it generates the same
`segmentation map reflected by apps coming out of the map decoder and
`applies that and sends the S to the inverse transform circuits.
`So, what the inverse transform circuit is doing is taking the incoming
`video information that comes through the transform decoder. It is
`segmenting that incoming video information into the identical block sizes
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`10
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`that were encoded at the transform coder of the encoder and then using those
`block sizes which correspond identically to the ones that are incoming from
`the encoder, it decodes.
`So, the segmentation on the reverse side is taking your incoming
`stream of video information and dividing it into the identical block sizes that
`were used during the encoding process. So, when you have an incoming
`video stream you're going to -- in the Kalker example it is your first coded
`syntax is a three, you're going to take an incoming video stream and you're
`going to take the first 16 by 16 bit and what the segmentation is taking your
`incoming video stream and taking it into those segments, 16 by 16.
`Now, the stream is continuing to come in and I may take the next one
`in an 8 by 8 segment and the one after that in a 4 by 4 segment or four sets
`of 4 by 4 segments. So, as the incoming video stream is being processed it's
`being segmented into chunks in a very and I apologize for the nontechnical
`term, but it is being segmented into chunks that will be processed in those
`chunks or blocks that correspond to the blocks or segmentation that was
`provided by the encoder.
`JUDGE ULLAGADDI: It is the Petitioner's --
`JUDGE MELVIN: Don't those S values represent just the size of a
`particular block and not a minimum or maximum size of the blocks in the
`image?
`MR. BURESH: The particular value as the stream of S's the 3’s, 2’ss
`and the 1’s are coming into the map encoder and those S values do represent
`the size of a particular block. In the petition we were very clear to
`distinguish, if you look at, for example, note 5 on page 29 of our petition.
`The same distinctions are drawn in the expert declaration.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`11
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`
`The value of S represents an individual block, but it is the syntax
`information that establishes the relationship between S and a particular block
`size and in the example in Kalker, the information that establishes that S
`equals 16 by 16, I'm sorry S equals three equals 16 by 16, is the syntax
`element they were pointing to for the second syntax element.
`The syntax information that is sent to the decoder to indicate that S
`equals one equals 4 by 4, is the first syntax element representing the smallest
`block.
`And that assignment information and is used to in Kalker generates a
`grid of the largest block size from which the system then operates, that is the
`syntax element that satisfies or syntax information that satisfies the first and
`second syntax elements. As I walk through this I can get into more detail. I
`understand that's based upon the questions and the institution decision, I
`understand that's a question and I'll continue to address that.
`Slide 14, quickly, just reiterates the segmentation is done at the
`encoder in Kalker by a segmentation circuit which is implemented into a
`segmentation map by the map encoder, it's based on a rate distortion
`analysis. The same basic analysis within the 365 patent.
`The video encoder scans in slide 15, the video encoder will scan based
`on the largest block size in the encoded unit. So, we note from this that that
`largest block size just like in the 365 is important and the way the largest
`block size is used if we turn to slide 16, and I want to note here, Your
`Honors, there are two embodiments in Kalker.
`And that the two embodiments are where Kalker will establish a grid,
`a segmentation grid based upon the smallest block size and then scan that
`grid based on the smallest block size and we are not relying on that
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`12
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`embodiment. I think you could argue whether it maps to these claims but
`we're not relying on that because there is a clear embodiment where you
`have Kalker generating the segmentation grid based on the largest block size
`and that is what I'm going to walkthrough.
`It's introduced here on slide 16 where you have three lines down and
`the quote first the top left of 16 by 16 block is analyzed and then the final
`sentence then the next 16 by 16 block is analyzed. So, from that we see that
`we're moving through the segmentation grid on a largest block size scale.
`We're moving 16 byblock by 16 by 16 block and that's further described on
`slide 17.
`Within those larger blocks of 16 by 16 as we proceed through the
`segmentation map you will also see this notion of dividing those larger
`starting blocks inset into smaller blocks, but for present purposes, what I'm
`trying to get across here is that the largest block size S equals three which
`represents 16 by 16, is used by the decoder to generate the segmentation
`map with a grid of the largest block size so that the processing can then
`occur on that largest block size grid from largest block to largest block to
`largest block if you proceed through a coded unit.
`JUDGE MELVIN: So, where does Kalker determine that the current
`block as its starting is equal to the maximum size?
`MR. BURESH: Yes. So, that will -- I have that claim on slide 21, if
`you turn there. Just in passing, as we go past slide 20 and I will answer your
`question. The slide 20 is just kind of to close the loop here and show that
`the segmentation information that is generated at the encoder is in fact
`transmitted to the decoder.
`It says here in Kalker, it must be transmitted to the receiver to allow it
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`13
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`to apply a backward transform to the same block as was used by the encoder.
`And that is just a preview. I include that to address the demonstrative that
`Patent Owner has brought forth in the reply which suggested that there are
`essentially different block sizes that could be somehow generated by the
`decoder itself.
`Slide 21 is determining that a block or a plurality of blocks and a
`sequence of pictures has a starting size equal to the maximum size using a
`second syntax element. That is the claim limitation Your Honor was
`referring to. And again, the way this operates in Kalker is that the largest
`block size in a particular coded unit is transmitted from the encoder to the
`decoder, which is the second syntax element.
`That second syntax element in the second embodiment in Kalker is
`used to establish a grid size at the decoder using the largest block size. And
`on slide 22, we see that the segmentation map is scanned on the basis of the
`largest block size. Now, in Kalker, if we turn back to -- go back to slide 18.
`That grid establishes the -- is used to determine the starting block size
`of the block. Now, if the starting block size in the claim limitation we're
`looking at here, is not necessarily the block size. It is just the starting block
`size. It may be the block size if the block that you're looking at, the current
`block, is at the largest block size.
`So, for instance, the upper left-hand block as you start your analysis in
`Kalker you assume it's a 16 byblock, so that is the starting size for your
`current block. You then assess the syntax information for that block and in
`this example the syntax information for the block is three. So, that current
`block is at the largest block size and no further syntax information is
`necessary.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`14
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`
`Then if you proceed to the next top right 16 by 16 block, so the
`starting size of that block was determined based upon the coded unit syntax
`information, the second syntax element that the starting size of that block
`would be 16 by 16. That block, however, the syntax information related to
`that particular block indicates that there is further sub partitioning.
`So, we need to go through and scan that block to determine what the
`subblocks are within that starting size of 16 by 16. So, the starting size for
`the current top right block was 16 by 16 based upon the coded unit syntax
`information establishing the grid for that starting size to be 16 by 16.
`From there, so you have current block has a starting size. The starting
`size of 16 by 16 and from there you go forth to a further partition based upon
`the syntax information indicating the partitioning of that particular block.
`So, Your Honor, that --
`JUDGE MELVIN: Does the Kalker code then send the size for every
`block?
`MR. BURESH: Kalker sends two categories of information. It sends
`coded unit information which would be the largest smallest and intermediate
`block sizes. In other words, it establishes the values of S equals three, the
`value of S equal two and the value of S equals one. So, that's one category
`and that's the coded unit which allows the decoder to establish the
`segmentation grid and also to understand the codes that it is going to be
`receiving for this particular coded unit.
`The second category of syntax information that Kalker transmits and
`receives is the individual partitioning information for the blocks and those
`would be the 3’s, 2’s, 1’s, that attach to a particular starting block or current
`block and then indicate the partitioning of that starting size current block
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`15
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`into a smaller block.
`JUDGE ULLAGADDI: Point number 2 where -- go ahead Judge
`Moore.
`JUDGE MOORE: I was just going to ask where in the decoding does
`Kalker use the size of every block?
`MR. BURESH: So, in the decoding in the process of recreating the
`segmentation map that was created at the encoder it utilizes the 3’s, 2’s that
`the streams of syntax elements that you see for example at column 5 and line
`60, the 3, 2, 2, 1, 1, 1, 2, 3, 3 in the sequence, that sets of syntax elements is
`what allows the decoder to create the segmentation map that is depicted in
`figure 9.
`So, you match up the syntax elements that indicate the particular
`partitioning of these blocks, the results is figure 9 and that figure 9 is what
`was generated by the encoder and transmitted through the two components
`of syntax information to allow the decoder to recreate that segmentation map
`that you're seeing in figure nine.
`JUDGE ULLAGADDI: Can you point us to where in Kalker it
`discloses that the mapping between S equals one or how it's transmitted or
`where it's transmitted, where in Kalker is this discussed?
`MR. BURESH: Sure. Yes. I'm going to come back to slide 23 and
`24 in direct response to your question, Your Honor. Turn to slide 25. Here
`the discussion, I think in the institution decision Your Honors asked a good
`question which is essentially how it -- when an ordinary skill in the art
`understands that the Kalker system is not hard coded such that no syntax
`information it sends or in order to establish that relationship if you will, S
`equals one equals 4 byand S equals three equals 16 by 16. So, the good
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`16
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`question is and the answer is twofold. I point to and have pointed to several
`things in Kalker that indicate to one of ordinary skill in the art that Kalker
`transmits syntax information to establish that value or that relationship. The
`first and I'm going to start from kind of the least critical to the most critical.
`Least critical in Kalker is a variable walk size system. I acknowledge
`that could be interpreted in two ways. Variable walk size within a particular
`coded unit versus that the encoder and decoder can communicate between
`coded units that there are different walk sizes at play. The second and again
`just kind of working on the priority scale, there are different examples of
`block size configurations in Kalker.
`So, the example we've spoken primarily on has the by 4 is the smallest
`and the 16 by 16 is the largest,but in column five you see another example
`where the smallest S equals one, this is lines 18 through 21 of column 5.
`Where S equals ones and notes 4 by 8 blocks and S equals three denotes 8
`by 8 blocks. Again, indicative, but not conclusive.
`Claim one, however, in fact, each is an independent claim has a
`concept in it. This concept is also in the summary of the invention app at
`column one, lines 45 through 50. If you look at and I refer to slide 25
`because it's easier for me to read. But slide has claim one which is an
`inquiry of those references I just provided to you.
`In the method of transmitting encoded video pictures of claim one in
`Kalker one of the last element here is transmitting the coded picture data and
`the coded segmentation map characterized in that the step of encoding said
`segmentation map comprises one, assigning a block size code to each block
`size and in two, scanning the segmentation map in accordance with a
`predetermined scan encoder.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`17
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`
`That step in there that I refer to as one of assigning a block size code
`to each block size and those block size if you look up in the first element of
`the claim you segment each picture into a map of picture blocks having
`variable block sizes. So, you have the block sizes established again based
`upon the rate distortion analysis.
`You then have this step of assigning a code. So, that's your S code to
`each block size that you just determined based upon your segmentation. So,
`that step of assigning a block size code to each block size is the opposite of a
`hardwired approach. You would not be assigning a block size code to a
`block size.
`In other words, I am seeing my segmentation into variable block sizes.
`I'm seeing that at the encoder and I'm saying I have a 16 by 16 block, I need
`to assign a block size code to it. Three is it becomes 16 by 16.
`Or I get my next picture image and the largest block size is 8 by 8.
`And that's another example from column 5. In -- that --
`JUDGE MOORE: Couldn't that just mean that for each block you're
`giving it a 1, 2 or 3 because where does it say that 1, 2 or 3 is different from
`picture to picture?
`MR. BURESH: That is the interpretation that Patent Owner presented
`and I think it's wrong for two reasons. One, it's not the words in Kalker. If
`you were assigning a block size code to each block I would agree with you,
`but that's not what's happening in the second step there. The scanning of the
`segmentation map in accordance with the predetermined scan order to obtain
`the one-dimensional series of block size code, that is where -- that is where
`you are assigning the same block size code to a particular block.
`That is different than establishing the relationship between the block
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`18
`
`
`
`IPR2019-00757
`Patent 9,930,365 B2
`
`size code and the block size. And that is what is called out in claim one and
`again also in the summary of the invention. That is establishing the
`relationship between a three and a block size, 16 by 16.
`Between a one and a block size, 4 by 4, that is not the same thing as
`establishing -- now three which I have already said to be 16 by 16 and -
`block. Those are two different concepts. And again, just the express
`language used here, and the express wording here is block sized code to
`block size. That is creating a relationship.
`JUDGE MOORE: Once again, Kalker's description that supports
`your reading of that language? And can you point us to where that is
`described?
`MR. BURESH: The -- again, just kind of walking backwards through
`what I've just gone through and again from the perspective of an ordinary --
`person of ordinary skill in the art when you are assigning a block size code
`for a block size you are doing that because you have variable block sizes in
`your various pictures.
`So, as you transition from one set of block sizes in one picture, one
`coded unit, to the next you are again you are working through assigning
`these values. If you didn't have -- if it was preordained that three equals 16
`