Judge Kozinski Reversed by Federal Circuit: Google May be Liable for Street View

By Dennis Crouch

Vederi v. Google (Fed. Cir. 2013)

This case arises from a decision by Alex Kozinski who is Chief Judge of the Ninth Circuit Court of Appeals who sat by designation in deciding on summary judgment that Google’s streetview product does not infringe Vederi’s asserted patents. U.S. Patent Nos. 7,239,760; 7,577,316; 7,805,025; and 7,813,596. Through his role on the Ninth Circuit, Judge Kozinski has played an important role in the development many areas of intellectual property and internet law. However, as with Judge Posner of the Seventh Circuit, Judge Kozinski has largely been denied any direct input in the patent context.  Here, that denial continues as the Federal Circuit has vacated Judge Kozinski’s decision — finding that Judge Kozinski erred in his claim construction that resulted in a non-infringement determination.

The patent here covers the method of mounting a set of cameras atop an automobile to capture images as the vehicle moves throughout a geographic area. Those images are then integrated with into a system that allows virtual visual navigation.

All of the asserted claims include a limitation that the views be “substantially elevations” Here, the term elevation appears derived from architectural usage meaning a flat representation of one façade of a structure and the Google Streetview images are better thought of as a perspective view. This difference can be seen somewhat by comparing Jefferson’s Rotunda sketch with a photograph of the Rotunda.


Judge Kozinski found that the Google Streetview images were not elevation views because they show images that are curved representation of the world.

On appeal, the Federal Circuit found that Kozinski’s view of something being substantially an “elevation” view was too limiting based upon the intrinsic evidence. In particular, the patent specification discloses the use of a fish-eye lens to create a photograph of the street in “substantially elevation” form. In the usual case, a fish-eye photo of a building will not be in elevation form but instead provide a curved representation of reality. The result then, according to the Federal Circuit, is that the “substantially” element of the claim provides substantial leeway on the meaning of elevation. The broader claim construction means that now Google may well infringe.

= = = =

An interesting element of the provisional priority document (filed in 2000) is that it includes a listing of the actual software code for taking the GPS and image data to create panorama views and allow a user to navigate through those results. The inventors here, Enrico Di Bernado and Luis Goncalves actually built the system back in 2000 while graduate students at Caltech by attaching a set of cameras and a GPS monitor atop Di Bernado’s car and driving around Pasadena. The results were then made available on the City of Pasadena website.

70 thoughts on “Judge Kozinski Reversed by Federal Circuit: Google May be Liable for Street View

  1. From the Hricik side of the blog, one wonders what Judge Kozinski is doing with any decision involving photography:

    Judge Kozinski was embarrassed, to say the least, when “private” images he
    stored (including undressed men appearing with sexually aroused animals) were
    not private, but publicly available because he did not understand the technology

  2. This is what happens when non-photography people talk about photography. They don’t think clearly.

    First of all, wide-angle is relative. Is a 50mm lens wide-angle? Is a 35mm lens wide angle? A 28mm? Does it have to have barrel distortion? If so how much? Is a 28mm without that rounding distortion a wide-angle lens?

    The reason I ask is that so much of the talk is about the fish-eye, which is a wide-angle lens (usually capturing 170 degrees plus in the larger dimension) that has serious distortion. The distortion is the fish-eye-ness.

    That distortion is readily mapped away. link to photo.net (among many sources)

    Here is the real problem: In the Jefferson rotunda photograph, the view of the dome is obstructed by the roof in the photograph (which will happen even with most telephoto shots made from below roof level). That is because the actual rotunda is a 3-dimensional object, and the roof building sticks out from the dome to get in the way of the view. The lens cannot fix that with some fancy math tricks. The lens is simply not viewing the missing part of the dome.

    The drawing is 2-dimensional. The building does not stick out to block the view of the dome, so the full dome can be seen from any angle of view of the drawing (except from behind). You see the full elevation no matter where you stand, no matter what lens you use.

    Any photographer of ordinary skill knows this stuff. Your best plan to get a full elevation shot is not to get a 50 foot tall lens, but to get at or above roof level and far away with a powerful telephoto lens. This will have the effect of flattening the building as much as humanly possible. Other than that, “substantially” is the best you are hoping for, especially from a vehicle on a street for multi-story buildings.

    1. Better prior art databases are not the priority (and are they really the issue ?). Priority should be given to efficient use of the already accessible databases by the Examiner of the USPTO. That alone could have a huge impact of the quality of the rejections.

  3. The CAFC panel unanimously rejects Kozinski’s claim construction. The new claim construction reads directly on reams of prior art.

    In fact, prosecution history reveals that most of the claims were rejected over §102 before “substantially elevations” was added. The remaining claim that didn’t require that limitation required the disclosed image stitching algorithm. That stitching algorithm is not the one Google uses and the ones Google does use are in the prior art. No problem for Google; they have the money to litigate over §102 and win handily.

    Of course, the old construction read on lots of prior art, also. The PTO resisted through the usual non-final rejection but eventually issued the patent. PTO should be ashamed of itself.

    I don’t see an IPR running over this. Will Google fight over §102 in court instead?

    None of this would matter if §101 were taken seriously; digital image processing and data retrieval are not patent eligible subject matter.

    1. digital image processing and data retrieval are not patent eligible subject matter.

      Why?

      I mean, other than because you feel like it.

      1. anon, I agree with you here. I don’t understand why even digital image processing is not eligible. It certainly was in Alappat.

    2. ’cause little Owen says it is not eligible.

      This is the problem with 101. You get ever APE sticking their tiny little minds into the picture.

      1. I don’t mind him sticking his mind in if he can back up his ‘opinion’ with something of merit, of substance. After all, the discussions here about the law are qualitatively different than academic discussions about the way we feel. This stuff is simply NOT things “subjective, completely in the mind and whatever we feel goes” kinds of stuff.

  4. OT – but more on waves (you know, those transient things) – evidence of transient things now being THE oldest things in the universe…

    (and yet, waves are still “too transient for patent judges….

    link to news.yahoo.com

  5. Here’s one of the claims:

    1. In a system including an image source and a user terminal having a screen and an input device, a method for enabling visual navigation of a geographic area from the user terminal, the method comprising:

    receiving a first user input specifying a first location in the geographic area;

    retrieving from the image source a first image associated with the first location, the image source providing a plurality of images depicting views of objects in the geographic area, the views being substantially elevations of the objects in the geographic area, wherein the images are associated with image frames acquired by an image recording device moving along a trajectory;

    receiving a second user input specifying a navigation direction relative to the first location in the geographic area;

    determining a second location based on the user specified navigation direction;

    and retrieving from the image source a second image associated with the second location.

    A couple immediate problems leap out regarding this junk. First, this is a claim to a computerized information storage and retrieval “system” that is perfectly identical to a gadzillion other storage and retrieval “systems” except for the information content being provided by the system.

    We have a broken patent system and this kind of junk being granted and asserted is Reason Numero Uno. Figure it out, people. The patent teabaggers will cry because they will have to do actually work instead of grifting off the broken system. So what? Screw them. Screw them to the wall and let ‘em dry up with their big fat mouths hanging open and their self-important walnut brains spilling out onto the sidewalk below.

    Second, let’s take a look at how the patentees here describe the prior art:

    The prior art further teaches the dense sampling of images of an object/scene to provide different views of the object/scene. The sampling is generally done in two dimensions either within a plane, or on the surface of an imaginary sphere surrounding the object/scene. Such a sampling, however, is computationally intensive and hence cumbersome and inefficient in terms of time and cost.

    Accordingly, there is a need for a system and method for creating a visual database of a comprehensive geographic area in a more time and cost efficient manner. Such a system should not require the reconstruction of 3D scene geometry nor the dense sampling of the locale in multiple dimensions

    Does anybody see anything in the above claim that gets around this prior art? Let everybody know.

    Lastly, I’ll note that from the viewpoint of computer “technology” (and at the level of abstraction in the claim here there is little or no “technology” to speak of) there is no patent-worthy distinction that can be drawn between this “technology” and the many, many, many first-person view video games and/or virtual navigation systems that allows you to virtually move through some defined space by presenting “images” of that space. None.

      1. your post uses the same old words that have been used a gazillon times before.

        Not unlike the claims. Receiving. Receiving. Determining. Receiving.

        What an unbelievable joke.

        Of course, there’s a big difference between my comments and “presumed valid” junk claims like these: you’re free to ignore my comments, but we’re all paying for these junk claims and our broken patent system, directly or indirectly.

        1. Of course, there’s a big difference between my comments and “presumed valid” junk claims like these

          Sure is – the big difference is that Malcolm’s junk comments are presumed invalid.

  6. Is this yet another example of how clients can spend millions on litigation attorneys without ever even getting to a patent validity evaluation, instead of also effectively using a vastly cheaper reexamination or [now] an IPR? Or is there a PTO delayed reexam decision on this patent and/or specific novel software details in the some of the claims?

    Speaking of 7th Circuit Judge Posner’s track record on appeal as a patent trial judge, isn’t it now zero for three with yet another likely reversal in the works at the Fed. Cir.?

    1. “Speaking of 7th Circuit Judge Posner’s track record on appeal as a patent trial judge, isn’t it now zero for three with yet another likely reversal in the works at the Fed. Cir.?”

      That speaks far more to the intelligence of the CAFC than Posner.

      1. Random Examiner,

        See posts 7.1 and 1.

        “Intelligence” per se is a smokescreen. Judges do not get to write patent law just because of their intelligence level.

    1. “It this is “substantially elevation” what is not?”

      It’s pretty blatant there is a 112 2nd implication to the language used in this patent. The line of demarcation between what infringes and what doesn’t is not clear.

      1. I have read just the first patent of the list, and the expression “substantially elevations” is not defined anywhere in the description; more precisely, the word “elevation” does not appear even once in the original application documents. From my perspective, it is baffling that so much time and money is spent to construe a term that was almost certainly “invented” by the attorney during the application’s prosecution.

        Right now I don’t have an opinion on whether the construction of the Fed. Cir. makes sense or not (I’d need to study the case more thoroughly), but it seems to me that a stricter standard on amendments in US prosecution would improve greatly legal certainty post-grant.

        1. the word “elevation” does not appear even once in the original application documents.

          Get. The. Eff. Out.

          Are you kidding?

          That explains why the Federal Circuit studiously avoids any discussion of the prosecution history, and where and when the term first appeared. They give lip service to Phillips and then turn to trashing the district court.

          1. MM, and never once point out that the term is not used in the spec.

            A superwide-angle view (fish eye) is not even close to being an “elevation” within the normal understanding of that term. To read the claim term to include what clearly and unambiguously it is not is exactly the problem you and I have discussed many times. There has to be a limit on construction of a claim term to mean exactly the opposite of what it ordinarily means.

            The claim term indeed is indefinite and the construction given the term by the Federal Circuit is bizarre.

          2. The opinion does say that the term appeared to overcome a prior art rejection; originally the claims cited “images providing a non-aerial view of the objects” (see p. 13 of the opinion), which was possibly unclear as well, but definitely broader in scope and thus anticipated. The opinion does not say, however, where the replacement term “substantially elevations” came from…

            1. the claims cited “images providing a non-aerial view of the objects” (see p. 13 of the opinion), which was possibly unclear as well, but definitely broader in scope and thus anticipated

              For all practical purposes, it’s pretty much the same breadth as the construction of the disputed term by the Federal Circuit, which includes “substantially front or side views” of objects. Other than “aerial” views of the object, what’s missing that anybody cares about? “Sub-surface views”, I guess.

              Again, I am left to wonder: how can it be the case that a method of stitching together photos of a subset of views of an object, claimed at that level of abstraction, can possible be patent-worthy over methods of stitching together photos of all views of an object? On what basis? Certainly not “unexpected results”, that’s for sure.

              It’s almost as if the PTO and the Federal Circuit likes to pretend that they were born yesterday as soon as the word “computer” appears in a claim. They come across as that pathetic and st00pit. When is this crybaby “art” of “computer technology” ever going to grow up?

        2. “Right now I don’t have an opinion on whether the construction of the Fed. Cir. makes sense or not (I’d need to study the case more thoroughly), but it seems to me that a stricter standard on amendments in US prosecution would improve greatly legal certainty post-grant.”

          That and reminding examiners of what 112 1st is every other week and reminding them that there should be a decent number of 112 1st rejections going out if they’re average. So many people here never do them. I can assure you that they’re missing 112 1st problems if they never do 112 1st rejections.

    2. As Google points out in its brief, car mounted cameras cannot take elevation shots of buildings; there must be fish-eye distortions even with the most flat of lenses. The only exception is when the lens is literally larger than the building itself.

      So CAFC is right to interpret substantially elevations as meaning photos of the front of buildings that face the street.

      The problem with the patent is §102 (and §101, of course), not unclear claim language or spherical image processing.

      1. Owen, I think “substantially elevation” must mean the images taken by a conventional camera having a fixed pointing direction relative to the movement of the camera in a given direction. Because a camera is used, the picture will not be a perfect elevation. But “elevation” to include super-wide angle views seems to read “elevation” out of the claim.

    3. what is not?

      A couple of things immediately come to mind:
      A purely plan view.
      A purely elevation view (which is exactly what Google errantly argued). Which by the way in a summary judgment stage should not prevail.

      1. A purely plan view.

        Right, just like the sub-surface views. That’s trivial and not relevant to either party (the patentee already tried to disclaim aerial views).

        A purely elevation view

        Nothing in the Federal Circuit’s construction excludes a “purely elevation view”. That would be a “front view”. Expressly within the construction provided by the Federal Circuit.

        1. You make the mistake that “trivial and relevant to either party” does not impact what the word means, and is somehow not pertinent. That the argument made by Google is wrong is very relevant and not so trivial.

          You then make the mistake of thinking the purely elevation view is some type of “inclusive” condition, when the “purely” modifier makes it an “exclusive” condition.

          You really svck at this Malcolm.

  7. It does not matter what the case is, if a famous judge from another circuit sits by designation in a district court patent case, the Federal Circuit will find a way to reverse his decision, merits be damned. I’m not a huge fan of Judge Kozinski generally, but he has more intellectual horsepower than the entire bench of the Federal Circuit combined.

    1. Lode_Runner,

      See post 1. I think you confuse “merits” with “proper role of the judiciary.”

      And further, while “intellectual horsepower” can be a great asset – it clearly can be misused when the hubris of the judge overwhelms his appointed station.

    2. But all that intellectual horsepower did not have sufficient strength to save the genius judge from a rookie blunder: conflating a term of art – an elevational view – with an obviously made-up patent term (“substantially elevations”) that requires construction, not unduly puzzling, in light of the specification (which dooms the conflation).

    3. Also, the decision was unanimous despite being issued by a panel of oil and water (Rader and Dyk). Which indicates pretty clear error.

      Your comment about “intellectual horsepower” brings to mind fond reminiscences of The Princess Bride, when Inigo mildly tells Vizzini that “you keep using that word. I do not think it means what you think it means.”

  8. All of the asserted claims include a limitation that the views be “substantially elevations” Here, the term elevation appears derived from architectural usage meaning a flat representation of one façade of a structure and the Google Streetview images are better thought of as a perspective view.

    I haven’t seen the claims but how in the world could this distinction possibly make a difference for patentability? i.e., why would anyone bother to put such a limitation in every claim unless they had a good reason?

    Don’t get me started on claims that broadly cover methods of taking pictures using an auto-mounted camera and making those pictures accessible electronically. That’s just a bad joke. But quite reflective of the sad, pathetic, broken state of our patent system generally.

    1. Interesting – an expression of the effect is removed, while the offending post causing the effect remains…

    2. MM, the only stitching disclosed was of regular camera shots as one progressed in a straight line. The pictures were not strictly elevations, no camera picture is.

      But wide angle is quite different. They are not substantially elevations unless one intends to read out of the claim “elevation.” To the extent Rader and his merry crew construed the term to cover fish eye views…. well once again I do not understand just how that is possible.

      Given the stakes, this one has to be take en banc or go up to the SC.

      1. Ned, I’m not even getting into the “details” of this silly claim construction.

        I’m just wondering why, if you think you are the awesome patent-worthy inventor of the method of putting a camera on a car and taking pictures of stuff for people to access electronically, why do you limit yourself to any kind of lens?

        Was there prior art to be avoided? And if so, how on earth could the choice of one old lens over another old lens make any difference whatsoever?

        That’s my question.

        1. “I’m just wondering why, if you think you are the awesome patent-worthy inventor of the method of putting a camera on a car and taking pictures of stuff for people to access electronically, why do you limit yourself to any kind of lens?”

          That was my thinking as well. I’m guessing that the PTO rejected “putting a camera on a car and taking pictures of stuff for people to access electronically” sans the special lens. Or they only filed on that since they figured the PTO would reject such ivo prior art they knew about.

          1. So the lens is the key element to patentability here and they somehow get away with use of not only the term “substantially” to describe that lens, but apparently a definition of the term so broad as to include lenses no skilled artisan would ever consider to fall within the term.

            And let me take a wild guess: the Federal Circuit’s “construction” of this claim nowhere informs anyone what sort of lenses/images their construction does not cover.

            1. Indeed, here’s the Fed Circuit’s “claim construction”:

              the record shows that “views being substantially elevations of the objects” refers to “front and side views of the objects.” Thus, as properly construed, the claims do not exclude curved or
              spherical images depicting views that are substantially front or side views of the objects in the geographic area.

              Note: the court provides no description of what sort of images are not included within the term “substantially elevations of the objects”. All we know is that the images of the fronts of objects and/or images of the sides of objects are included within the scope. Given that, I ask again: why in the world would a patentee choose to use its bizarre “subsantially elevations” language in its claims unless it was (i) insane or (ii) trying to deceive people (e.g., the PTO)? Why not just say “substantially front or side views” so the breadth was clear to everyone?

              And if that’s the construction, then what is all this business in the specification about overcoming the “computationally difficult” problems in the prior art relating to “3-D scene geometry” or “dense sampling”? It appears that the “solution” to this “problem” that these geniuses came up with is (wait for it) take less pictures. Really? That’s worthy of patent? “The prior art reference is capable of doing way more than what we’re doing, and therefore requires more computation.” Seroiusly?

              Check this out from the spec:

              these statements in the specification [about the prior art] refer to the dense sampling of an object using cameras placed in a sphere surrounding the object looking inwards. Thus, the 3D scene geometry being discussed is using the photographs surrounding an object to reconstruct its 3D shape. In other words, this reference is fundamentally different from using a cluster of cameras to take images from a particular point looking out in all directions.

              “Fundamentally different”? In the context of stitching together images? That’s pathetic. The prior art teaches that any overlapping images of any object can be stitched together with a computer. Given that fact and the generically hand-wavey “technology” described in the claims, there is no “fundamental difference” between a computer that stitches together images from 20 cameras surrounding a fire hydrant on all sides and a computer that stitches together two elevation images of a dog. It’s just stitching images together. Conceptually — which is the level of abstraction presented in the claims here — the technology is absolutely identical.

              Oh but wait — dogs and fire hydrants are “fundamentally different” in this context, right? I’m sure the Federal Circuit thinks so and I know the USPTO thinks so.

              We have a broken patent system.

          2. “but apparently a definition of the term so broad as to include lenses no skilled artisan would ever consider to fall within the term.”

            That’s what it is appearing at the current moment.

            “the Federal Circuit’s “construction” of this claim nowhere informs anyone what sort of lenses/images their construction does not cover.”

            Of that you can be pretty much assured.

        2. MM, you are right that the only type of image excluded are the admitted prior art of two- and three-dimensional views, and areal views. The kind of lens is now immaterial, and it makes no difference that a perspective view taken by a wide-angle lens is not normally considered to be an elevation view.

          I just cannot get over the fact that the federal circuit gave the ordinary meaning of elevation no weight at all. The claim term was added to distinguish the prior art. If the term did so in a way that clearly and unambiguously excluded a disclosed embodiment, even if unnecessary to distinguish the prior art, they should be held to it.

          I can’t get over the feeling that there was a miscarriage of justice here.

  9. Will Chief Judge Kozinski leave his perch on the Ninth Circuit and sit by designation on the remand?

  10. I am not a camera buff, but I think the difference between the patent and Google might lie in the use by Google of a wide-angle lens. The patent discloses two kinds of lenses, a normal lens and a fish eye. A fish eye lens is, I think, a super-wide angled lens. Whether this would produce a “substantially elevation” is really questionable. I think Google’s argument that the term itself would exclude wide-angle lens, and was intended to claim photos using conventional lenses that were not precisely elevations.

    This is an interesting case on just what one can claim given the disclosure of two species of lenses – whether it can cover yet a third. Are they equivalent? I do not know. But what I do know is that I agree with the District Court.

      1. you’ll know I often avoid the normal rules of grammar, spelling, and punctuation.

        GASP – Oh, the horror !!!

        /sardonic bemusement off

      2. “I often avoid the normal rules of grammar, spelling, and punctuation”

        I’m not sure “avoid” is the exact word you meant there.

        1. not sure…

          Why would you think so?

          Clearly, the thought being expressed is that the good professor is choosing not to take the time to express his thoughts and views according to the normal rules (choosing to be informal and quick rather than formal and slow).

          This is neither surprising nor unbecoming.

          In fact, I have coined a phrase dealing with this very type of intended decision: finding the right tree in the wrong forest.

          The take away is to focus on the content and not the exacting grammatical rules.

          Now if the professor had said “you don’t need to be honest,” then THAT would have deserved a comment like yours.

  11. What the heck; we’re giving Europe to the Russians, so we may as well give our technology to the Chinese by scrapping the patent system. We can always go back to subsistence farming like the North Koreans.

    1. give our technology to the Chinese by scrapping the patent system.

      Can you please explain the mechanics of how scrapping the system whereby we publish enabling disclosures of all of our technology will allow “the Chinese” to get our technology?

      1. The system of reverse engineering would allow them to build the systems and the system of preventing the copying in the US would be gone.

        1. But they’ll still have our technology if we’re publishing it right? I think that was the point of the comment, not that we’ll lose the protectionism for the american market.

      2. It is not the publishing that is the problem, IANAE…

        …Well, in a way it is – it is the publishing (taking the Quo) and not giving any Quid that is the problem.

  12. Should judges have such a role?

    re: “Through his role on the Ninth Circuit, Judge Kozinski has played an important role in the development many areas of intellectual property and internet law. However, as with Judge Posner of the Seventh Circuit, Judge Kozinski has largely been denied any direct input in the patent context.

    This statement seems to assume that common law development in the patent sphere is a good thing – let alone a thing allowed by our constitution.

    How deep the rabbit hole of “implicit?”
    When does “development’ of the law become a de facto writing of the law?

    (btw, it appears that the sixth paragraph currently ending in “since none of the embodiments“” has been cut off from the rest of the sentence/paragraph)

    1. What these judges want to do is write patent law by making findings of fact. Such as Posner’s finding of fact that inventors just are like that and like to invent. They don’t need incentive according to Posner. They like to do it. He once shared a dorm room with an inventor and knows. Ergo, we need no patents.

      1. Perhaps some academic out there would like to research why the U.S. Constitution did not place the authority to write patent law in the judiciary…

      2. “What these judges want to do is write patent law by making findings of fact.”

        Sounds suspiciously a lot like what judge “a compooter that is reprogrammed is a new compooter” Newman did in Alappat. Obviously whether x is anticipated by y that is a question of fact. Determining that x is not anticipated by y is also a determination of fact. According to anontroll et al. that is somehow “the law” as opposed to “a finding of fact”.

          1. Well, there you go projecting again.

            Sorry 6 – that “all in your head subjective say whatever you want” thing is, ….

            … all in your head.

Comments are closed.