Jim Bell Q'n'a, Questions Please Options

[djellison]

, but since HiRISE is apparently using Jpeg2000, I would guess that MER could as well. I wonder whether the compression scheme in MER over

For the raw quick-look releases (like we get with MER at the Exp and JPL) HiRise will be J2K, but the calibrated published data down the line will be uncompressed IMG's or similar I'm fairly sure.

Doug

[ugordan]

As I understand it the compression used prior to transmitting the files is a type of wavelet compression (which is good), however why they opted to "wing it" versus using an ISO standard is what I'm wondering about. My guess is that the standard wasn't finalized at the time they were building the hardware, but it's just a guess.

The main reason for developing a proprietary compression algorithm was better performace in regards to data transmission packet sizes and error resilience, both specific to the medium the data are transmitted along - radio waves over the DSN network. It was heavily influenced by Jpeg2000, only adapted to a very specific use. I have a PDF describing the MER algorithm somewhere, but can't seem to dig it up now.

I wonder whether the compression scheme in MER overlaps with that in Jpeg2000 so that a direct conversion would be possible without any loss of quality.

While both standards are very similar in terms of compression principles, I don't believe they're that compatible. There are specifics to each algorithm that probably make a lossless conversion impossible.

[algorimancer]

For the raw quick-look releases (like we get with MER at the Exp and JPL) HiRise will be J2K, but the calibrated published data down the line will be uncompressed IMG's or similar I'm fairly sure.

Doug

Actually it looks like you were correct the first time. I did some googling yesterday and came up with a justification to the PDS as to why Jpeg2000 was suitable (things like high compression quality and stable ISO standard, as well as the subsampling capability (you can pull a lowres thumbnail out of a j2k image without having to read or process the the entire image file). I gather it was approved and is being used. However it wouldn't surprise me if they downsampled/converted to regular Jpeg format for immediate release, while later posting to PDS in the original j2k format. Incidentally, while j2k has a lossless compression option, it would surprise me if they used it routinely. I gather, for instance, that MER occasionally returns lossless images, particularly with the MI, but otherwise some compression is routinely used. In my experience 99% quality compression is virtually indiscernable from lossless, and a fraction of the size ... meaning we get lot's more returned images than otherwise.

[algorimancer]

I have a PDF describing the MER algorithm somewhere, but can't seem to dig it up now.
While both standards are very similar in terms of compression principles, I don't believe they're that compatible. There are specifics to each algorithm that probably make a lossless conversion impossible.

This may well be that PDF, which describes the ICER compression scheme used by MER and compares it with j2k and others:

http://tmo.jpl.nasa.gov/progress_report/42-155/155J.pdf

As I read it, it rather sounds like it is using something like a subset of capabilities of j2k. While it is clear that a non-lossless j2k image could be losslessly converted to the ICER format, the reverse is not obvious.

[djellison]

You average compression for 'earth' use isnt going to be robust enough to withstand a the odd bit error, so I image they do use a subset of j2k, but just modified for robustnes s:)

Doug

[ugordan]

This may well be that PDF, which describes the ICER compression scheme used by MER and compares it with j2k and others:

http://tmo.jpl.nasa.gov/progress_report/42-155/155J.pdf

Yes, that's exactly the one. Thanks for digging it up!

As I read it, it rather sounds like it is using something like a subset of capabilities of j2k. While it is clear that a non-lossless j2k image could be losslessly converted to the ICER format, the reverse is not obvious.

If the reverse is not possible, then the whole thing is useless in terms of releasing the PDS raw data as any recompression will make it less raw, and obviously not any different than regular JPEG. That's not to say it's impossible to repackage the bits and re-encode into J2000, as long as the wavelet coefficients aren't touched themselves.

[djellison]

You have till 1700UT today for new questions - I'm doing another session with Jim later

Doug

[Toma B]

Just one simple question...
When will Opportunity reach Victoria Crater...
I know that nobody knows that for sure but I would like to hear from someone as competent as Jim Bell is...
Thanks!
BTW, Doug, did you solve the problems with audio recorder...remember the last time...

[djellison]

That's one of those questions that's not worth asking, like the end of the Steve Q'n'A - I actually said I wouldnt ask how long they'll last - it's just one of those things you don't know.

When will Oppy reach Victoria Crater is like, how long's a piece of string.

I could ask how agressive they intend to be with driving, what sort of thing it would take to flag up a science stop, how the rover is in terms of health - But I don't think anyone knows how long it'll actually take.

I've tested Skype, and it's sounding much better now Still not perfect, but a lot better ( and infinitely cheaper ) than a phone call

Things I intend to cover, some odd Pancam issues ( the white pixels down each side, the frames that are just noise etc ) - how was the resolving power of Pancam chosen - and the development of Pancam from way back in the pathfinder days.

And of course, their current status

Doug

[Toma B]

I could ask how aggressive they intend to be with driving, what sort of thing it would take to flag up a science stop, how the rover is in terms of health

Well, I’m not trying to be offensive but, if Jim Bell does know how aggressive they intend to drive and what sort of thing it would take to flag up a science stop, as you sad it then he CAN give approximate date when they will arrive...
Thank you anyway!

But I don't think anyone knows how long it'll actually take.

Just his opinion...
P.S.

I actually said I wouldnt ask how long they'll last

That is whole other question. I think it would be unethical to ask that...

[algorimancer]

As mentioned previously in this thread, I would be really interested to hear his take on the Jpeg2000 question, noting that HIRISE is apparently using it, and why they went with the proprietary ICER equivalent wavelet compression method instead. For that matter, could they send up a software patch and change to the more portable compression standard even now? Leading up to, it would be really nice to be able to download and view the raw MER images without the requirement of Nasa software.

[djellison]

it would be really nice to be able to download and view the raw MER images without the requirement of Nasa software.

img2png



Doug

[algorimancer]

img2png



Doug

Or for that matter, img2bmp or img2tiff, the problem is that while there is no loss in quality, the compression is lost (losing the benefit of the small file size), along with the possibility of various wavelet analyses which can be extracted from the original wavelet-based image; there is a LOT of interesting information in the raw data of a wavelet compression.

[djellison]

You've lost me. The very best data is the RAD .img files as far as I know - what is it you're actually after instead of that?

Doug

[centsworth_II]

You could just invite Jim Bell to join the arrival sol at Victoria lottery.