US 2013/0272429, a patent application on calculating three checksums instead of one

Question

I stumbled upon patent US 2013/0272429 that appears to patent calculating three checksums instead of one.

Even if the obviousness of the basic idea is not enough to throw this out there should be plenty of prior art available for calculating checksums for each color plane in various old picture and video compression standards, container standards and open source software.

The rest of the patent seems to be HEVC specific, and boils down to "to calculate the checksum for the image, you have to decode the image like it's HEVC".

Answer 1

Even if the obviousness of the basic idea is not enough to throw this out there should be plenty of prior art available [...]

There is no such thing as "obviousness of the basic idea" in the absence of relevant prior art. I do not mean to say that this specific patent application is necessarily non-obvious, just that that's not how you determine obviousness in the context of patent law.

[T]here should be plenty of prior art available for calculating checksums for each color plane in various old picture and video compression standards, container standards and open source software [...]

That may be, but I wouldn't bet on it. Why would you want to calculate a separate checksum for each color plane? Certainly, it's likely that some lossless image compression scheme can be found that involved calculating a checksum for a whole image to ensure proper decoding. But why do it separately for each channel?

Indeed, I'd say there's no clear utility for the patent, and that might be a good rationale for attacking it without looking too hard at the prior art.

The specification does mention that "[t]his checksum may be used by the decoder for compliance testing, bit stream transmission or storage error detection, etc" but it is questionable whether that is a specific utility. Any other method of transmitting checksums that covers all of the data achieves these goals equally well.

Answer 2

This is not a patent. It's a publication of an application for a patent.

As for prior art, a quick search comes up with US20120307151 A1, which discloses calculating a checksum for the entire 24-bit YCbCr value. There may be closer references.

FYI, "obviousness" is a far stricter standard than you imply. Obviousness generally refers to a straightforward combination of prior art that would have been realized by a person of ordinary skill in the appropriate discipline -- for example, if one reference discloses performing checksum validations on 8-bit Red, Green and Blue values individually and another reference discloses performing a checksum validation on a 24-bit YCbCr value as a whole, a claim to performing 8-bit Y, Cb, and Cr values individually might be deemed "obvious" in view of the combination of aforementioned references.