Jim Bell Q'n'a, Questions Please Options

[djellison]

OOops - too late

I got the JPEG2K question in, asked about the cloud pictures, the Spirit dragging, how agressive they intend to drive - will they stop for the dark cobbles or the baby craters, picking filters for end-of-drive imaging, and the history of pancam right back to the pushbroom jobbie in the early '90s.

AND....news just in. Odyssey safed yesterday, so no pretty pictures for a day or two.

Doug

[algorimancer]

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

Fundamentally I'm after the best possible quality images in an international standard format. If the image is lossless, that would mean either PNG or Jpeg2000 (which has a lossless mode), though I'd settle for Tiff or Bmp (and happily convert to one of the others). If the image is compressed (most are), then wavelet compression yields the best quality with fewest artifacts (and is what they use on MER), and Jpeg2000 is the standard format. Admittedly Jpeg2000 involves additional processing overhead, but when dealing with limited bandwidth they seem the optimum solution.

There are just all sorts of benefits to wavelet-compressed images, completely aside from file size. Upon display they yield minimal artifacts, scale smoothly, and you can even pull an arbitrary resolution thumbnail out of one without having to read the entire file. Aside from that, there are a slew of papers out there on such topics as pattern recognition and texture analysis simply based upon the wavelet coeficients (similar things are possible with Jpeg images). I'm happy that they're using wavelet compression to begin with, I'm concerned that by choosing a non-standard format they have to some extent isolated the potential data analysis to within the JPL community.

My understanding of the .img file structure is that it contains a lot of useful header data plus a tag specifying the raw format of the image data, and then the actual image data is in the originally acquired format. For instance, the Hughens probe sent back Jpegs, so within the corresponding .img files there should be that raw Jpeg data, and likewise for the MER images in terms of the ICER format.

Anyway, I see that you've already done the interview and I look forward to hearing what he had to say

[djellison]

My understanding of the .img file structure is that it contains a lot of useful header data plus a tag specifying the raw format of the image data, and then the actual image data is in the originally acquired format. For instance, the Hughens probe sent back Jpegs, so within the corresponding .img files there should be that raw Jpeg data, and likewise for the MER images in terms of the ICER format.

That's not my understanding of it - I think the imagery data within an IMG is totally uncompressed (thus every full frame IMG is always 2mb. If they were in any way compressed, they would vary)

Doug

[helvick]

That's not my understanding of it - I think the imagery data within an IMG is totally uncompressed (thus every full frame IMG is always 2mb. If they were in any way compressed, they would vary)

But some of them are rebuilt from an original that was compressed, right? like the 1bpp, 4bpp etc images?

[djellison]

Yes - a compressed image comes down, and they then rebuild a calibrated image from that, and all the other settings related to the camera at that time. Flatfield, Darkfield, Smear, temperature etc etc - to produced something calibrated.

Doug

[algorimancer]

I stand corrected Surely they must archive the original raw images somewhere...

[djellison]

Well - they must do somewhere, but the 'product' is something made from that data, not the data itself.

Doug

[Toma B]

May I ask when (or where) will yesterday's Jim Bell's Q'n'a be posted?

[djellison]

B)-->

May I ask when (or where) will yesterday's Jim Bell's Q'n'a be posted? [/quote]

The usual place ( http://www.planetary.org/explore/topics/ma...vers/audio.html ) - and once Emily's had a chance to whack it up there, she'll post a thread in this sub-forum to say so

Doug

[edstrick]

Bear in mind that the rovers have a 20 MhZ CPU, I believe. I think an important factor in their selection of compression methods is computing time required.

[djellison]

Bear in mind that the rovers have a 20 MhZ CPU, I believe. I think an important factor in their selection of compression methods is computing time required.

Going on what Jim said yesterday - the most important thing is robustness. Your compression has to be able to withstand a lost bit or a lost packet, and not trash the whole image - hence why they developed something themselves. It's also highly tweakable and can compress things like the sky very very well, but still retain good detail in things like layering etc.

Doug

[odave]

For the next time around - How often has autonav been used, percentage wise, and how well do they feel the autonav system has performed so far?

[djellison]

See - I'd TOTALLY forget if you didn't put in here. I'm useless at things like that.

Things I want to get to next time include - autonav and visodom
Albedo measurements - what they actually achieve
How much of a mission could there be sans-Odyssey

and as an aside, CTX and MARCI performance with their test images.

and anything else you lot come up with

Doug

[alan]

I'd like to know how much Spirit's stuck wheel will impair its ability to climb steep slopes. Of course with the varying traction I don't know if I should expect a very specific answer.

[Pertinax]

Looking back in time a little bit: Has there been any sucess in creating an aprox. true color image of Spirit Sol 675's Phobos eclipse while Phobos was in Mars' umbra? If so, anything that can be shared at present?

-- Pertinax