AIM Sport and Supponor have been fighting in Germany, the UK and now at the UPC over whether Supponor’s TV sports advertising image processing system infringes AIM’s EP 3295663 B1. Supponor have since been re-badged as TGI Sport but the battle goes on.
The score to date:
- In Germany, the patent was first found infringed, but this was reversed on appeal by the Munich Higher Regional Court. In separate validity proceedings the patent was deemed valid.
- In the UK, at first instance the patent was found valid and infringed. On appeal, the infringement decision was upheld but the patent was only considered valid after amendment.
- In the UPC, the patent was found valid but not infringed.
Here, we’re going to see how and why the UPC Local Division interpreted the claim so that there was no infringement, compared with the approach taken by the UK courts lading to a different result.
What’s the invention?
The claim at issue was independent method claim 12 and to be fair it looks a bit fierce and hard to understand. It’s reproduced below with the key parts underlined, and then explained further below. Helpfully, the different decisions reviewed here used the same feature numbering.
12 A method of digitally overlaying an image with another image,
12.1 comprising creating (200) a model of a real world space,
12.1.1 wherein the model includes an overlay surface to be overlaid with an overlay image,
12.1.1.1 wherein the overlay surface in the model represents a display device in the real world,
12.1.1.2 wherein the display device is configured to display a moving image on the display device in the real world by emitting radiation in one or more predetermined frequency ranges;
12.2 identifying (201) camera parameters, which calibrate at least one camera with respect to coordinates of the model;
12.3 capturing (202) at least one image with respective said at least one camera substantially at the same time, said at least one captured image comprising a detection image,
12.3.1 wherein the camera used to capture the detection image is configured to detect radiation having a frequency outside all of the one or more predetermined frequency ranges and distinguish the detected radiation outside all of the one or more pre-determined frequency ranges from radiation inside the one or more pre-determined frequency ranges;
12.4 positioning (203) the overlay surface within said at least one captured image based on the model and the camera parameters;
12.5 detecting (204) an occluding object at least partially occluding the overlay surface in a selected captured image of said at least one captured image based on an image property of the occluding object and the detection image;
12.6 overlaying (205) a non-occluded portion of the overlay surface in the selected captured image with the overlay image, by overlaying the moving image displayed on the display device in the real world with the overlay image in the selected captured image.
To present the invention in more friendly terms, consider its main implementation at a football match. A TV camera points at the pitch and sees the LED advertising boards arranged around the pitch. The system is calibrated so that it knows exactly where these advertising boards are located. In real life, the advertising boards display moving images, and these are generated at narrow frequency ranges associated with the red, green and blue LEDs. The TV camera captures an image of the pitch, players, ball and advertising boards. What the system wants to do is to be able to present different moving images exactly overlaying the advertising boards, for different broadcast markets. But of course, no-one wants the overlaid advertising image also to cover the view of the players on the TV screen.
So what the method does is to produce a “detection image” in which the frequency of light from the advertising boards is filtered out. In that filtered detection image, the advertising boards just look dark and blank. The system knows where the advertising boards are (due to the calibration) and so the system is looking for “occluding objects” (i.e. football players) in front of the advertising boards in the filtered detection image. Then, the broadcast image can be generated to include the overlaid advertising image only onto the non-occluded parts of the advertising board.
The key point for claim interpretation was at feature 12.5. What does “image property” mean? Does it include just checking the brightness of individual pixels in the filtered detection image? Or does it require something more sophisticated?
Helsinki LD – literal interpretation
The outcome in AIM Sport v. TGI Sport (29 April 2026) turned on the interpretation of this feature 12.5. The court noted that prior art systems aimed to detect the non-occluded parts of the advertising board based on an infrared signal included in the light from the advertising board. So, for pixels where this signal was present, that was non-occluded (“light”) and where this signal was absent, that was occluded (“dark”). In the jargon used by the parties, this was “dark on light” detection of occluding objects. The approach used in the patent was instead “light on dark” detection – i.e. seeing the player in front of a dark advertising board in the filtered detection image.
According to the court, the description of the patent gave three “definitions” of the image property of the occluding object. It may be fairer to say that these are presented as examples, particularly as they map onto definitions presented in the dependent system claims. These examples were of relatively complex approaches: (i) considering the differences between neighbouring pixels; (ii) the differences between stereo images; and (iii) changes in the images over time. The court appears to have taken the view that (ii) and (iii) were not inside the scope of claim 12, despite being the focus of separate dependent system claims in the granted patent.
The court expressed their interpretation of feature 12.5 like this:
Hence, for the person skilled in the art, the image property of the occluding object can be construed at best as based on an intrinsic property (or characteristic) of the object which can be identified from the detected image. Such a characteristic makes it possible to distinguish occluding objects from the overlay surface in the detected image.
They also specifically stated:
The brightness of pixels in the detection image alone cannot be regarded as an intrinsic image property of an occluding object since the brightness value of a pixel depends not only upon the reflectivity of this object at the detected radiation frequency but also upon an external factor, namely the intensity of the ambient light illuminating the object at this frequency.
On validity, the court agreed with the patentee that the idea of detecting the player rather than the advertising board was a “paradigm shift”, and the finding of inventive step was based on this being something that the skilled person would not have done starting from the prior art.
The alleged infringement was the SVB System of TGI. This had a basic mode of operation that simply did “dark on light” detection of a first infrared frequency which was not argued to infringe. However, it had a supplementary mode for bright ambient light conditions which took into account the brightness of a pixel at a player compared with a darker pixel corresponding to the advertising board, at a different infrared frequency. So this could be thought of as “light on dark”. Relying on their comments about claim interpretation, the mere brightness of a pixel corresponding to a player was not considered to be an intrinsic image property of the player. So there was no literal infringement. This outcome was the same as at the first instance appeal in German national litigation.
Helsinki LD – doctrine of equivalents
AIM argued that there was infringement by equivalence. Both sides argued this point based on the four step test set out by The Hague LD in Plant-e v. Arkyne (22 November 2024). Although equivalence has come up a fair bit at the UPC, no other Local Division has either set out an alternative test, or explicitly endorsed the one from The Hague LD. However, as we explained in UPC Weekly 2025 w24, UPC decisions on this topic have explained that there will be no infringement if there is no “technical-functional equivalence” of the variant.
The court in this case said they would follow The Hague LD test because both parties had advocated for it and because there was no UPC Court of Appeal guidance suggesting something different.
In the view of the Helsinki LD, there was no infringement in view of the first question:
Technical equivalence: does the variation solve (essentially) the same problem that the patented invention solves and perform (essentially) the same function in this context?
The court noted that the invention was not the overlaying of images onto the non-occluded part – this was in the prior art. The invention was the use of an image property of the occluding object rather than in effect detecting the absence of a signal corresponding to the advertising board. This seems fair. They then turned to the question of “context”, deciding that the context of the patent is different to the SVB System because the patent aims to work with regular advertising boards rather than ones that are modified with infrared LEDs. On whether the variant performs essentially the same function, the court decided that:
The function of feature 12.5 is not found to be present in the SVB System since the latter does not detect an occluding object based on an image property describing an intrinsic characteristic thereof.
And therefore there was no technical equivalence. It is interesting that in the application of this part of The Hague LD DoE, the court focused on whether the variant provided the same function as the element of the patent claim, but expressed this function not using the wording of this part of the claim or the overall effect of the feature but as the court’s narrower interpretation of it (the “intrinsic characteristic”).
UK Court of Appeal – infringement on normal interpretation
At first instance in AIM Sport v. Supponor (EWHC 164 (Pat) 30 January 2023), the UK Patents Court held that the patent as granted was valid and infringed. The appeal judgment in AIM Sport v. Supponor (EWCA Civ 396 23 April 2024) was that the patent as granted was invalid, but valid when amended to include the requirement at the end of the claim that the advertising board is a uniform monotone in the detection image as if it was not active.
Both the Patents Court and the Court of Appeal agreed that the granted claim includes pixel brightness as an “image property”. The judges at both instances noted the discussion of higher order processing examples in the description, and the fact that there were no examples that disclosed pixel-by-pixel processing. However, this could not override the view that the term “image property” in the claim is broad – it is a property of an image:
Limitations which are not present in the claim language are not to be read in by reference to examples which appear in the specification. The skilled person reading a patent understands that the examples are simply that. If the claim language is broad – as the term “image property” clearly is – then the claim is correspondingly broad, for good or ill.
The Court of Appeal also commented on whether the claim should be limited to what it termed the “inventive concept”. This was based on Supponor’s argument that the inventive concept disclosed in the patent relied on higher order processing than mere pixel-by-pixel processing. The view of the Court of Appeal was that it was only possible to identify the inventive concept after properly interpreting the claims, and so the inventive concept would not be the use of higher order processing.
The result was that the SVB System was found to infringe the patent as granted in view of its use of the pixel brightness at the player location in the supplementary mode of operation.
Where next?
Proceedings in the UK are over. We understand that the German proceedings are still pending via a request to appeal to the German Federal Court of Justice. In view of the approach of the parties to national litigation and the value of the case, it seems possible that there will be an appeal to the UPC Court of Appeal. Their view on literal infringement will be interesting to contrast with the first instance. It’s also possible that they will have to consider infringement by equivalence, which would be big news indeed. We will watch and report again.
-1.png)
.jpeg)