[00:00:00] Speaker 01:
All right, our next case for argument today is 23-1634, Bank of America versus Nance Holden.

[00:00:08] Speaker 01:
All right, come on up.

[00:00:09] Speaker 01:
How do I say your name, counsel?

[00:00:11] Speaker 02:
Emmerich Rage-Clessis, your honor.

[00:00:13] Speaker 01:
Mr. Rage-Clessis, please proceed.

[00:00:15] Speaker 02:
Good morning, your honors, and may it please the court.

[00:00:17] Speaker 02:
I'd like to start with the three cases that turn on the combination

[00:00:21] Speaker 02:
of Ogaswara and Boll where the board made two key legal errors.

[00:00:26] Speaker 02:
First, despite its assumption that Boll's technique would produce better results than Ogaswara's software, which was a rational motivation to combine, the board found the combination non-obvious because the embodiments in Boll are stationary and therefore physically different from Ogaswara's mobile device.

[00:00:47] Speaker 02:
And that was legal error, because the motivation inquiry is not limited to the primary uses or physical embodiments of the prior art.

[00:00:56] Speaker 02:
Second, the board legally erred in requiring proof that a camera flash was not only known, but well known in the art.

[00:01:04] Speaker 02:
No such legal burden exists.

[00:01:07] Speaker 02:
None of the claims here even recite a flash.

[00:01:10] Speaker 02:
And a camera flash is undisputedly part of the bowl, which was part of it.

[00:01:14] Speaker 01:
And just to be clear, do you have to win all of these arguments in order to have us send it back?

[00:01:18] Speaker 01:
Are these independent bases, or do you have to win them all?

[00:01:21] Speaker 02:
Your Honor, we have to win on two bases for the three proceedings that involve the combination.

[00:01:27] Speaker 02:
The issue of what I'll call movement, in terms of whether Bull's technique could be applied to a mobile device, and the issue of the flash, which the board limited to reasonable expectation of success.

[00:01:38] Speaker 02:
With those two issues, we win on those three proceedings, Your Honor.

[00:01:45] Speaker 02:
And a camera flash, in any event, is undisputedly disclosed by Bull, which was part of our combination.

[00:01:51] Speaker 02:
And in fact, the patent owner's own expert here admitted that a flash would have been known to persons of skill and art.

[00:01:57] Speaker 01:
Well, whether there is a motivation to combine, is that a question of law or fact?

[00:02:00] Speaker 02:
In general, Your Honor, it's a question of fact.

[00:02:02] Speaker 01:
But sometimes it's not?

[00:02:04] Speaker 01:
I mean, by in general.

[00:02:05] Speaker 01:
I don't know, the qualifier doesn't.

[00:02:06] Speaker 02:
It is a question of fact, but whether the board applied the proper legal standard in evaluating the facts in the record is a question of law.

[00:02:15] Speaker 02:
And there has to be substantial evidence that supports the board's result under the proper legal standard.

[00:02:21] Speaker 01:
So if you're suggesting the board applied the incorrect legal standard, what sentence in the board's opinion would you point me to that shows that they failed to appreciate what the motivation to combine test was?

[00:02:32] Speaker 02:
Well, I would point first to, I think it's clearest in the 004 patent proceeding, Your Honor, particularly at appendix 190.

[00:02:41] Speaker 02:
Because in that proceeding, the board acknowledged- 190?

[00:02:45] Speaker 02:
190, Your Honor.

[00:02:46] Speaker 01:
So I'm looking for a statement of law that's incorrect.

[00:02:49] Speaker 01:
So where is the statement of law that's an incorrect statement of law?

[00:02:52] Speaker 02:
To be clear, Your Honor, it's not a statement of law.

[00:02:55] Speaker 02:
Let me be clear.

[00:02:56] Speaker 02:
We're not contending that the board misstated the legal standard.

[00:03:00] Speaker 02:
We're contending that in the way the board applied its analysis of motivation and reasonable expectation of success, it did so improperly such that- Well, that's facts.

[00:03:10] Speaker 01:
That's fact-finding.

[00:03:12] Speaker 01:
It is fact-finding, but the substantial- But a minute ago, you told me this was a legal question.

[00:03:16] Speaker 02:
The reason I think it's a legal question, Your Honor, is that even if we accept the board's factual premises here, they actually support a motivation to combine under the legal standard.

[00:03:26] Speaker 02:
And I think it's simple because the board assumed, for purposes of its decisions to start out, that there was a rational reason to combine these references.

[00:03:34] Speaker 02:
It assumed our motivation that Boll's technique would produce better results.

[00:03:39] Speaker 02:
That's in Appendix 44.

[00:03:41] Speaker 02:
112 and 191, just for all three of those proceedings.

[00:03:45] Speaker 02:
It also assumed that Boll had the processing, excuse me, Ogasuara's videophone had the processing power to run that technique.

[00:03:52] Speaker 02:
That's at appendix 61, 127, and 198.

[00:03:56] Speaker 02:
There's no finding of teaching away, and there's no dispute that these references are analogous art.

[00:04:01] Speaker 02:
So that gets us to the issue of movement.

[00:04:03] Speaker 02:
where the board found that pixel perfect registration was necessary at the first step of Boll's method.

[00:04:10] Speaker 02:
But even if we assume that, Boll tells us that there are two ways to achieve that pixel registration for segmentation.

[00:04:20] Speaker 02:
One is to have a stationary scene where the object and background are stationary relative to

[00:04:25] Speaker 02:
the camera.

[00:04:26] Speaker 01:
And the other is to... What a piece of prior art discloses is also a question of fact, correct?

[00:04:30] Speaker 02:
That is correct, Your Honor.

[00:04:31] Speaker 01:
So in this case, the board found, as a factual matter, that Bowles is stationary for a critical reason associated with its segmentation method.

[00:04:40] Speaker 01:
Is that correct?

[00:04:41] Speaker 02:
That is correct, Your Honor.

[00:04:43] Speaker 01:
So why did the board err in light of that?

[00:04:48] Speaker 02:
Well, the board acknowledged at that page 190 that I mentioned that there is a quick succession teaching in Bull, which states that Bull's camera may be stationary.

[00:05:00] Speaker 02:
But the technique that Bull provides allows for some motion in the sense that if you take the images in quick succession, one right after the other, then any motion is insignificant or irrelevant to the segmentation.

[00:05:14] Speaker 02:
It does not prevent spatial registration from being sufficient to segment the images at the first step of Bull's method.

[00:05:24] Speaker 01:
I'm sorry.

[00:05:24] Speaker 01:
Didn't the board find that that quick succession method was for the stationary camera?

[00:05:29] Speaker 02:
It did, Your Honor, and that's because Bull... I don't understand.

[00:05:34] Speaker 01:
It's like you want me to just unravel all these fact-finding games.

[00:05:38] Speaker 01:
Almost de novo, it seems.

[00:05:40] Speaker 02:
And we're not asking that, Your Honor.

[00:05:42] Speaker 02:
We will accept the board's finding that Boll relates to a stationary camera.

[00:05:47] Speaker 02:
But the fact that Boll's inventors didn't think that a CCD video camera in this method could be applied to a mobile device doesn't mean it was not obvious to do so.

[00:05:57] Speaker 02:
The question is whether Ogasuara's device, which is also a video CCD camera, same type of device,

[00:06:04] Speaker 02:
is taking images in quick succession.

[00:06:06] Speaker 02:
It's a handheld camera, though, right?

[00:06:08] Speaker 02:
It is a handheld camera.

[00:06:09] Speaker 03:
That was the concern, that here we need some kind of essentially pixel-by-pixel accuracy when we're taking two images with a bold stationary camera.

[00:06:21] Speaker 03:
And then the question is, could you really do some kind of pixel-by-pixel accuracy if you were to take two successive photos with a handheld camera?

[00:06:33] Speaker 03:
And the board said,

[00:06:34] Speaker 03:
No, they didn't think based on the evidence that that would be a smart thing to try to expect shoppers at Target holding one of these handheld devices to expect that they'd be able to take two pictures in a row, or that's not what you would want with that kind of a device.

[00:06:51] Speaker 03:
And so it wouldn't make sense, therefore, to do that kind of modification to Agassawara's mobile handheld bar scanner or whatever.

[00:07:01] Speaker 02:
I think that's a fair summary, Your Honor, but that raises the question of whether Ogaswara's device is taking images in quick succession, which Boll explicitly teaches is an alternative to maintaining a stationary scene.

[00:07:14] Speaker 03:
One with a flash and one without a flash.

[00:07:16] Speaker 03:
Correct, Your Honor.

[00:07:17] Speaker 03:
So now we have a customer at Target, it's in an aisle, looking at a product, trying to take two pictures of the product, one with a flash, one without a flash, and we're expecting that the customer will be able to hold that camera super still so that we're going to have that kind of pixel by pixel accuracy between the two images?

[00:07:42] Speaker 02:
Well, just to be clear, Your Honor, it's not two separate

[00:07:45] Speaker 02:
pictures, it's taking a video.

[00:07:47] Speaker 02:
So the board assumed that Ogaswara's video phone is taking a video.

[00:07:51] Speaker 02:
So it's taking frame by frame video.

[00:07:52] Speaker 02:
There's no serious dispute about that.

[00:07:55] Speaker 02:
In fact, the board acknowledged that Ogaswara is taking a video at appendix 56, 124, and 193.

[00:08:02] Speaker 02:
which it found requires taking images, quote, in a short period of time.

[00:08:06] Speaker 02:
So that's quick succession.

[00:08:08] Speaker 02:
The board did not analyze, however, in its discussion of motivation to combine whether Ogaswara satisfies the in-quick succession embodiment in bowl, which is taught as an explicit alternative to maintaining a fully stationary scene.

[00:08:23] Speaker 01:
I feel like you're just pointing to other evidence in a way that the board could have made its fact finding.

[00:08:27] Speaker 01:
But you seem to be completely ignoring the fact that we review this for substantial evidence.

[00:08:32] Speaker 01:
And the evidence in this record is that each, the bold disclosure is that pixels are being compared and they need to be in the same location.

[00:08:42] Speaker 01:
They have to be stationary.

[00:08:43] Speaker 01:
That's the fact finding the board made.

[00:08:46] Speaker 01:
And that is why there's no motivation to combine this with, you know, somebody moving around at Target with a camera judge.

[00:08:52] Speaker 01:
Jen had, you know, this is not, it's got to be stationary.

[00:08:55] Speaker 01:
So it's a fact finding.

[00:08:57] Speaker 02:
Right, the fact finding is that Bull requires pixel-perfect registration.

[00:09:01] Speaker 02:
I understand that.

[00:09:02] Speaker 05:
Can I ask you something?

[00:09:03] Speaker 05:
Sure.

[00:09:03] Speaker 05:
What is the standard of review that we apply to that?

[00:09:06] Speaker 02:
The standard of review for fact findings is substantial evidence.

[00:09:09] Speaker 05:
And what does that mean?

[00:09:10] Speaker 05:
I mean, for all the students in the room.

[00:09:12] Speaker 02:
Substantial evidence means that there's sufficient evidence in the record that a reasonable fact finder could have arrived at the result.

[00:09:20] Speaker 02:
But it has to be under the right legal standard.

[00:09:22] Speaker 02:
And it has to be substantial evidence that supports the result under the right legal standard.

[00:09:27] Speaker 05:
My problem is that you keep on saying under the right legal standard, but I don't see where it was an incorrect legal standard.

[00:09:33] Speaker 05:
So assuming for a minute that it wasn't an incorrect legal standard, then what is your argument on substantial evidence?

[00:09:41] Speaker 05:
Do you have one?

[00:09:42] Speaker 02:
Because if we accept the board's findings that number one, Bull requires pixel-perfect registration and that quick succession is still requiring that pixel-perfect registration,

[00:09:55] Speaker 02:
It just begs the question of whether Ogaswara's device is taking images in quick succession.

[00:10:00] Speaker 02:
And that's something the board didn't analyze.

[00:10:03] Speaker 02:
And so that's the problem here.

[00:10:05] Speaker 02:
There's a disconnect in terms of the board's analysis.

[00:10:08] Speaker 02:
It simply found, well, we need pixel perfect registration.

[00:10:11] Speaker 04:
So your primary reference, what does your primary reference teach?

[00:10:16] Speaker 02:
The primary reference Ogaswara teaches a mobile phone which has a CCD video camera.

[00:10:22] Speaker 02:
It is taking frame by frame video images one right after the other.

[00:10:27] Speaker 05:
And you want to modify that in view of bowl to teach what?

[00:10:33] Speaker 02:
We want to modify Ogaswara by applying an object recognition technique that is in bold, which also relies on the same type of device, a video CCD camera.

[00:10:44] Speaker 05:
But it says it's important that it be stationary, or it does not disclose an environment in which its technique would work.

[00:10:52] Speaker 05:
with a mobile device, right?

[00:10:54] Speaker 02:
Of course, and I think then we'd be in the realm of anticipation.

[00:10:57] Speaker 05:
And so what the board said, that it basically didn't think it would have been obvious to make this modification that you're proposing because it doesn't make sense where you need something that's

[00:11:08] Speaker 05:
pixel-to-pixel registration for where somebody with a camera, a cell phone, something somebody's holding, it would be a very difficult thing to do.

[00:11:16] Speaker 05:
So it seems like a difficult combination, right?

[00:11:20] Speaker 05:
A person with ordinary skill wouldn't look too bold to modify glycerin.

[00:11:25] Speaker 02:
I think that's true in terms of the board crediting some testimony that this was non-trivial and prone to be erroneous.

[00:11:32] Speaker 02:
But that doesn't necessarily mean that it wasn't obvious to combine, particularly when the board- But there's contrary evidence.

[00:11:38] Speaker 05:
There are expert testimony and rationale for why a person who ordered a nurse here on the yard wouldn't have been motivated to make that modification.

[00:11:46] Speaker 02:
Well, I'm not sure.

[00:11:48] Speaker 02:
I would say that there is expert testimony that's relevant to that question.

[00:11:52] Speaker 02:
But the board never weighed the testimony from the patent owner saying that this would be non-trivial or potentially more difficult to do on a mobile device.

[00:12:01] Speaker 03:
You thought your expert's testimony would be conclusory, right?

[00:12:04] Speaker 03:
That's what the board's finding was?

[00:12:06] Speaker 02:
There is, in two of the proceedings, there is a relative credibility finding, only in two of them.

[00:12:12] Speaker 02:
But to be clear, we don't need our expert's testimony on this particular issue, because there's actually no dispute, even with the patent owner's expert, that you can have some pixel errors here, and it would still work.

[00:12:24] Speaker 02:
And his testimony is just that it would be nontrivial and prone to be erroneous to put this on a mobile device.

[00:12:29] Speaker 03:
You're running out of time.

[00:12:30] Speaker 03:
Do you want to just say a few seconds about the 252 patent?

[00:12:34] Speaker 02:
Yes, Your Honor.

[00:12:34] Speaker 02:
I think for the 252 patent, the issue is very quick.

[00:12:39] Speaker 02:
The issue is simply whether Ogaswara teaches the determining limitation of claim 18.

[00:12:44] Speaker 02:
The board made a simple mistake that its opinion turned on.

[00:12:47] Speaker 02:
At appendix 217, the board found that Ogaswara discloses that the user of the device verifies that an image contains an entire symbol when capturing the image.

[00:12:58] Speaker 02:
And therefore, it concluded that Ogaswara couldn't possibly make that determination.

[00:13:01] Speaker 02:
But the quote that the board cites for that proposition does not say that.

[00:13:06] Speaker 02:
It says, the customer can visually verify the entire image has been captured.

[00:13:11] Speaker 02:
And that was essentially the crux of what led the board astray and led it to ignore the explicit disclosures in Ogaswara itself that show that it does two steps in terms of recognizing and decoding.

[00:13:24] Speaker 02:
And with that, unless there's further questions, I'll save my time.

[00:13:27] Speaker 01:
OK, Mr. Glass.

[00:13:28] Speaker 02:
Thank you.

[00:13:35] Speaker 00:
Thank you.

[00:13:37] Speaker 00:
It's a privilege to be here today, and if it may please the court, I'll start exactly where counsel left off.

[00:13:42] Speaker 00:
The board didn't misquote Ogaswara.

[00:13:44] Speaker 00:
He left off a pretty important quote.

[00:13:46] Speaker 00:
The board went on to say that, as part of the program, the customer need only inform the system that the image is ready for processing by pressing a predefined key on the system's keyboard.

[00:13:57] Speaker 00:
That's what Ogaswara is about.

[00:13:58] Speaker 00:
You have a handheld scanner.

[00:14:00] Speaker 00:
You press a button.

[00:14:01] Speaker 00:
You tell the system it's ready for processing.

[00:14:03] Speaker 00:
There's no determination stage.

[00:14:05] Speaker 00:
That's at Ogoswara, appendix 1470, column 20, line 47.

[00:14:11] Speaker 00:
I'll stay on Ogoswara for a minute.

[00:14:13] Speaker 00:
In the briefs, appellant argued that, well, we need to look at claim 12.

[00:14:18] Speaker 00:
We argue this is a new claim.

[00:14:19] Speaker 00:
This is a new argument.

[00:14:20] Speaker 00:
They said, no, it's just new evidence for the same argument we're raising below.

[00:14:25] Speaker 00:
It's not.

[00:14:26] Speaker 00:
You're talking about the 252 patent?

[00:14:28] Speaker 00:
The anticipation patent, your argument, 252.

[00:14:30] Speaker 00:
OK.

[00:14:30] Speaker 00:
Correct.

[00:14:31] Speaker 00:
Claim 12 requires recognizing and translating.

[00:14:36] Speaker 00:
Below, they don't focus exclusively on recognizing and decoding.

[00:14:41] Speaker 00:
August 4 is very clear that decoding and translating are two separate steps.

[00:14:48] Speaker 00:
Decoding, I won't belabor the issue because this wasn't briefed below.

[00:14:52] Speaker 00:
Decoding is taking a barcode and just generating a number.

[00:14:55] Speaker 00:
As there's no meaning behind that number.

[00:14:57] Speaker 00:
That's an August war.

[00:14:58] Speaker 01:
Why don't you move on to the motion to combine argument if that's okay.

[00:15:01] Speaker 01:
Absolutely.

[00:15:03] Speaker 00:
I'll start where they started.

[00:15:06] Speaker 00:
One moment on this notion that the board

[00:15:11] Speaker 00:
expressly found that there were better results from the combination.

[00:15:15] Speaker 00:
That's one of that has become by the gray roof.

[00:15:17] Speaker 00:
That became one of their primary arguments.

[00:15:20] Speaker 00:
The board made no such rule.

[00:15:22] Speaker 00:
In fact, the board found exactly the opposite.

[00:15:24] Speaker 00:
Council cited to Appendix 43.

[00:15:26] Speaker 00:
The board specifically stated

[00:15:29] Speaker 00:
We find this proposed combination of teachings to simply not make sense based on the record before us and the relevant time frame.

[00:15:37] Speaker 00:
That's where they get that the board found expressly assumed better results.

[00:15:43] Speaker 05:
Is this at page A, 43?

[00:15:44] Speaker 00:
This is at appendix 43.

[00:15:47] Speaker 05:
OK, and what part of it are you looking at?

[00:15:50] Speaker 05:
I'm sorry.

[00:15:51] Speaker 00:
I have my notes on this, but it starts with, we find, I think it's at the bottom of 43.

[00:15:56] Speaker 05:
We find no credible explanation in the record before us.

[00:15:59] Speaker 05:
Is that the sentence?

[00:16:00] Speaker 00:
We find this proposed combinations of teachings to simply not make sense, even based on the record before us.

[00:16:08] Speaker 00:
So if the board found that the combination doesn't make sense, how could they have expressly assumed and expressly have found

[00:16:18] Speaker 00:
that the combination would have yielded better results.

[00:16:20] Speaker 00:
They didn't.

[00:16:21] Speaker 00:
What counsel pins their argument on is the next sentence where they say, even assuming that ball segmentation methods yield better results in object feature recognition, as petitioner asserts, that's not an assumption they made.

[00:16:34] Speaker 00:
They said your combination doesn't make sense.

[00:16:37] Speaker 00:
So how could there be better results if the combination itself doesn't make sense?

[00:16:42] Speaker 00:
Counsel complained that there was no

[00:16:46] Speaker 00:
that the board discounted their expert's testimony that there was a relative weighing.

[00:16:51] Speaker 00:
The board went much further than that.

[00:16:53] Speaker 00:
I'm reading from Appendix 42.

[00:16:54] Speaker 00:
This is in the 1081 proceeding.

[00:16:57] Speaker 00:
The board found we find Dr. Bajaj was our expert.

[00:17:01] Speaker 00:
Dr. Rodriguez was their expert.

[00:17:02] Speaker 00:
We find Dr. Bajaj's testimony on this issue much more credible than Dr. Rodriguez.

[00:17:08] Speaker 00:
which we find to be misaligned at best with Bahl's explicit disclosures.

[00:17:13] Speaker 00:
Every argument they're raising today on whether there's alignment permitted in Bahl, these were all raised below.

[00:17:20] Speaker 00:
Their expert raised these exact arguments below, and the board found them not credible.

[00:17:26] Speaker 00:
The board actually found them misaligned with the disclosure of Bahl.

[00:17:34] Speaker 00:
Your Honor, just on what

[00:17:36] Speaker 00:
Ball expressly requires, Ball could not be clear that the pixels, I won't belabor this too much, but Ball is very clear that the pixels have to be in alignment.

[00:17:48] Speaker 00:
I know your honors are familiar with the record.

[00:17:50] Speaker 00:
Ball takes one picture without a flash, another picture with a flash.

[00:17:54] Speaker 00:
It uses what I referred to below as a brute force method of segmenting an object like an apple or produce from the background.

[00:18:02] Speaker 00:
And it goes one by one.

[00:18:03] Speaker 00:
It looks at one pixel from one picture, another pixel from the other picture.

[00:18:08] Speaker 00:
It looks for differences in intensity from the reflection of the flash and segments out those pixels.

[00:18:14] Speaker 05:
So what you're talking about right now, just for the students, you're identifying the substantial evidence to support the fact-finding.

[00:18:21] Speaker 05:
that Boll requires a pixel-to-pixel registration.

[00:18:24] Speaker 00:
Is that right?

[00:18:24] Speaker 00:
You put it better than I did, yes.

[00:18:26] Speaker 00:
That's exactly what I'm pointing to.

[00:18:29] Speaker 00:
We cited two pages of this in Boll.

[00:18:32] Speaker 00:
I'll just read what I think is the best statement.

[00:18:35] Speaker 00:
This is at Appendix 1498.

[00:18:37] Speaker 00:
The pixels in each pair being compared must correspond to one another, i.e., being the same respective location in each image.

[00:18:47] Speaker 00:
That's what the board based its decision on.

[00:18:49] Speaker 00:
That's what we argued below.

[00:18:51] Speaker 03:
At some point, didn't the board say something about how bowl does tolerate some small amount of movement between the two images?

[00:19:01] Speaker 03:
Yes.

[00:19:01] Speaker 03:
So now we're kind of in a messy area where the board acknowledges that ball allows for some movement, and yet made some conclusion that, well, whatever that amount of movement is OK with that camera.

[00:19:21] Speaker 03:
it doesn't translate over to Ogasawara's camera.

[00:19:27] Speaker 00:
I think it's a little more straightforward than that, Your Honor.

[00:19:31] Speaker 00:
What the board found was that, just to be clear, their proposed combination takes Ball's segmentation technique unaltered and drops it into Ogasawara.

[00:19:43] Speaker 00:
What the board found was that Ball is very clear that spatial registration is required.

[00:19:49] Speaker 00:
So you can, and this is what led to this one.

[00:19:51] Speaker 03:
But you can get spatial registration apparently with some small movement in the camera.

[00:19:57] Speaker 00:
As long as the pixels maintain alignment.

[00:20:00] Speaker 00:
So that's what led to this, I argue this below, that's what led to this question from the board.

[00:20:06] Speaker 03:
The pixels are pretty small, right?

[00:20:08] Speaker 00:
Exactly.

[00:20:08] Speaker 03:
So the small movement that the board was confessing Bull can do?

[00:20:14] Speaker 03:
It's not just small.

[00:20:15] Speaker 03:
It has to be miniscule.

[00:20:16] Speaker 03:
In totesimal.

[00:20:17] Speaker 03:
Exactly.

[00:20:18] Speaker 03:
Vanishingly small, not just small.

[00:20:21] Speaker 03:
The board didn't say vanishingly small.

[00:20:23] Speaker 03:
I'm just trying to understand what the board was trying to get at.

[00:20:26] Speaker 00:
But that's the gist of their decision.

[00:20:28] Speaker 00:
And they focus on the notion that Ball states, well, you can have spatial registration by, this is Appendix 1498, at column 10, line 41.

[00:20:39] Speaker 00:
Proper registration can be ensured by either acquiring the first and second image in quick succession,

[00:20:44] Speaker 00:
or taking them separately.

[00:20:47] Speaker 00:
That doesn't change the fact that Ball expressly states the pixels must be aligned.

[00:20:52] Speaker 00:
And all the board was saying was, yes, you can take them quick, but you've got to take them really quick so that you maintain that pixel registration.

[00:21:01] Speaker 00:
This is what led to this one millimeter issue that's in the record.

[00:21:05] Speaker 00:
The board asked me, I explained this to the board below, they asked me, well, how big is a pixel?

[00:21:10] Speaker 00:
I said, I don't know.

[00:21:10] Speaker 00:
I have no idea how big a pixel was in 2000.

[00:21:13] Speaker 00:
I said, I guess one millimeter.

[00:21:15] Speaker 00:
And I pointed out, that's their burden.

[00:21:17] Speaker 00:
Ball is objectively clear that the pixels must be spatially registered.

[00:21:22] Speaker 00:
If they're going to argue that there must be some movement, that there's allowed to be some movement, they should have put in evidence.

[00:21:29] Speaker 00:
And they complained that we, I said on the fly during a hearing that, well, I guess maybe about 1 millimeter.

[00:21:37] Speaker 00:
And if you look at the board's ruling, the board never used that 1 millimeter as part of their ruling.

[00:21:42] Speaker 00:
They stated, for example, they said on the order of 1 millimeter.

[00:21:47] Speaker 00:
Their ruling was based on Ball's express statement that spatial registration was required.

[00:21:55] Speaker 01:
Okay, do you have anything else, counsel?

[00:21:57] Speaker 00:
I have one quick point.

[00:21:58] Speaker 00:
I'll make this very brief.

[00:21:59] Speaker 00:
On this alleged admission that our expert made, I would just urge your honors to look.

[00:22:03] Speaker 00:
This is that Appendix 2873.

[00:22:05] Speaker 00:
It spans four pages.

[00:22:08] Speaker 00:
If you look at what they quote, they take three sentences from our expert that he admitted that spatial registration wasn't required.

[00:22:16] Speaker 00:
They take one quote, one sentence from the top of one page.

[00:22:19] Speaker 00:
It's separated then by, I think, 25 or 30 lines.

[00:22:22] Speaker 00:
They take another quote from the bottom of that page.

[00:22:25] Speaker 00:
No ellipses, no indication to the court that they're combining these sentences artificially.

[00:22:30] Speaker 00:
If you look at what he testified to repeatedly, he said, that is not how Ball works.

[00:22:36] Speaker 00:
That's not my testimony.

[00:22:38] Speaker 00:
Ball specifically says you need to have them in perfect registration.

[00:22:42] Speaker 00:
I won't belabor that point.

[00:22:44] Speaker 00:
I just suggest the honors, because that is a big issue in their briefs.

[00:22:47] Speaker 00:
I would just suggest reading that testimony.

[00:22:49] Speaker 00:
I think it's instructive.

[00:22:50] Speaker 01:
Okay, thank you, Council.

[00:22:52] Speaker 00:
Thank you, Your Honors.

[00:23:00] Speaker 02:
Thank you.

[00:23:01] Speaker 02:
And briefly, Your Honor, I do want to touch on something that Judge Chen raised, which is there is an inconsistency in the factual assumptions across these decisions.

[00:23:09] Speaker 02:
Because in the 004 patent proceeding, the board explicitly found that Boll's quick succession embodiment allows for small unintended movement, including for the camera, including when the support for the camera is not firm.

[00:23:24] Speaker 02:
Whereas in the 529 and 036 patent cases,

[00:23:27] Speaker 02:
it assumed that you couldn't have even a pixel's worth of movement, which it quantified as not even a millimeter or less.

[00:23:34] Speaker 02:
So there is an inconsistency across the factual assumptions between these board decisions.

[00:23:39] Speaker 02:
In the 004 patent case, the board went on to find that that embodiment is for a stationary camera only.

[00:23:45] Speaker 02:
But there's no principled reason and no discussion or evidence in the record as to why it should make a difference that it is the object or background that is moving as opposed to the camera.

[00:23:56] Speaker 02:
If you have an object moving away from a camera or a camera moving towards an object,

[00:24:01] Speaker 02:
there is no principle distinction for why that should matter.

[00:24:04] Speaker 02:
It shouldn't matter that Boll's inventors didn't think that you could apply their technique to a mobile device.

[00:24:10] Speaker 03:
Ogaswara came out a few years later and put this... But Boll says that its segmentation occurs on a pixel-by-pixel basis, right?

[00:24:19] Speaker 03:
Correct, Your Honor.

[00:24:19] Speaker 03:
Quick succession addresses it.

[00:24:21] Speaker 03:
So even if there's a finding that there's some little bit of movement that can occur,

[00:24:28] Speaker 03:
you still have to have, in order to have this facial recognition and segmentation, accuracy on a pixel-by-pixel basis.

[00:24:37] Speaker 02:
Two quick points on that, Your Honor.

[00:24:38] Speaker 02:
That's correct for the segmentation phase, but the quick succession embodiment explicitly states that you can get to that pixel-perfect registration by taking images in quick succession.

[00:24:49] Speaker 02:
And the second point, Your Honor, I see I'm out of time.

[00:24:52] Speaker 00:
I can finish my sentence.

[00:24:54] Speaker 02:
Is that this is all only about the first step of Bull, which is segmentation.

[00:24:58] Speaker 02:
There is no testimony or evidence that Bull just gives up and stops if there's a segmentation error.

[00:25:04] Speaker 02:
It keeps going and does its best at the object recognition phase, which merely requires a nearest neighbor classification or distance rule metric.

[00:25:13] Speaker 02:
And that is unaddressed by the board's opinions.

[00:25:15] Speaker 02:
Bull only needs a best match, not a perfect match, and so even

[00:25:18] Speaker 02:
small errors in segmentation would not preclude recognizing an object, which is the goal of the invention here.

[00:25:25] Speaker 01:
Okay, I thank both counsels.

[00:25:26] Speaker 01:
This case is taken under submission.

[00:25:28] Speaker 02:
Thank you, Your Honor.