Back to the Backshell

SUMMARY - I will present a super-resolved image of the Opportunity backshell/parachute and make a very initial analysis of the state of the backshell

INTRO - On Sol 66 of the Opportunity mission, the rover took 33 pictures of the backshell and parachute from a distance of about 1440 meters. I obtained the radiometrically calibrated IMG files and converted them to the lossless, large, but easy to manipulate BMP format. The original image sizes were 100x200 pixels.

IMAGE STACKER - In order to super resolve at the sub-pixel level (and not just merely improve the signal to noise ratio) then I needed to stack the images onto a finer background matrix. To do this I wrote a stacker in C++. I used one photo as the master that all others would be stacked against. I projected the master onto the finer backgound pixel grid that was nine times finer than the original image size; that is, the 100x200 pixel images [20,000 bytes] became 300x600 [180,000 bytes]).

I identified a 16x64 rectangle (1024 bytes) on the background projected master that contained high contrast (both bright and dark pixels; part of the backshell and part of the ground). The stacker moved the new image around the master and subtracted one image from the other in the ‘alignment rectangle’ area (we are left with 1024 ‘differences’). I took the mean of the 1024 differences, subtracted the mean from each of the 1024 diffs, and found the standard deviation of the whole rectangle (after each pixel by pixel nudge). The program registered each picture to the master by looking for the smallest standard deviation in that rectangle, 32 images were sub-pixel aligned in this manner.

I then used a cheap commercial deblur software package to tweak the soft edges (FixerLabs). The result are Images 1, 2, 3 (2 and 3 are just increased brightness, contrast)

FIRST APPROXIMATION ANALYSIS – The backshell is broken.

HiRise indicates that the backshell slammed onto the surface and created a persistent crater as seen from orbit. The backshell has a checkerboard pattern of light and dark areas. The dark areas are open seams of the backshell.

If you look at Opportunity’s ‘heatshield’ from HiRise it also has light areas with one dark stripe. From ground truth we know that is where the heatshield is broken open.

By contrast, Spirit’s backshell is homogenously light, there is no apparent or persistent impact crater and the parachute appears to have gently settled out of the sky in an inflated configuration. The Spirit backshell is not broken.

In the super resolved image of Opportunity’s backshell you can see ripped open seams (top left side of the backshell), that appear as dark points. The resulting open seam correlates to one of the seams on the HiRise image of the backshell.

In the super resolved image there is a faint circular pattern on the camera facing surface of the backshell. It appears to be either (1) a massive inward dent, (2) we are looking inside the backshell through another (larger) ripped seam, (3) this is debris from inside the backshell flopped outward (maybe the internal plate that holds the parachute canister).

Opportunity’s backshell in the HiRise image is distinctly ‘out of round’ (by contrast see Spirit’s backshell).

When you vertically exaggerate the super resolved image of Opportunity’s backshell, you will see the result of the impact on the hull. The triangular shape of the backshell should just be steeper but still remain distinctly triangular. The top of the hull has been broken open and the panels have been ripped outward. It is almost as if the circular plate near the top of the backshell (that holds the parachute canister and the lowering bridle for the rover) has been wrench out (or around) by the impact.

There is a small rock that has been super resolved, just to the right of the backshell. The parachute correlates well to the HiRise image (especially the bright white lump). The dark spots near the parachute are shadows or maybe rocks/debris thrown up by the impact. The dark areas appear as shadows in the HiRise image

The impact crater has also been resolved. It starts as the dark streak to the left of the backshell (which is not a long backshell shadow, since it is only 2:30 in the afternoon). The crater ends at the light triangular pattern under the left side of the backshell hull. That pattern is the crater wall.


Compare these images to the PanCam site super-res:
http://marswatch.astro.cornell.edu/pancam_...superres_4x.tif


Cheers!

Click to view attachmentClick to view attachmentClick to view attachment

That was really nicely done, and I particularly appreciate that you improved upon the "official" superres version . That crack is really obvious in your version (terrain behind it is clearly visible in the upper portion), not so in the other. This has the tone of a conference presentation (LPSC?).

OK, I'm impressed.

I love this idea and the results definitely look far superior to the official super-res. What do the traditional superres folks think? I was under the impression that the only way to get reliable signal to noise improvements was to build a process based on the point spread function of the camera.

Algorimancer: thanks but this is not a an LPSC paper it was something I cobbled together this winter waiting for baseball season

Helvick: the technique came to me while thinking about data reduction issues at work, but it must already be in use by somebody.

Noise can enter by stacking issues (creating blur) or image noise issues (stacking reduces that) and defocus (PSF). So all I did was stack carefully, do a general but light deblur and then contrast enhance.

the best of the bunch is SR_OPBS_IMG3

cheers

That's truly amazing...a very clever technique and very impressive results. Have you considered doing any of the other superres views?

Tom,

I will try to do some others.

I think I will start with the heatshield at Bonneville. That will also get me to make the tool more automatic vs the kludgyness of 3 programs running in a dos window now.

Stay Tuned.. and thanks

I think the assumption is that in the case of MERB - the residual RAD firing at bridal cut flung the backshell and chute in a westerly direction, but the chute did not re-inflate. They followed an arc, impacting at very high speed, thus breaking the backshell and creating quite an impact. Whereas, if you look at the HiRISE imagery of both MERA and Pathfinder, you can see that the backshell is sat, upright, on the surface, with the chute to one side suggestive of the chute reinflating (at least partially) between that residual RAD impulse, and impact.

GREAT results - there were a LOT of super-res images taken while Opportunity was parked up with IDD trouble around Olympia (West end of Erebus) - if you're looking for food for your process, that's the place to go


Doug

I was looking at the Cornell pancam superres page to see if I could choose a favorite to nominate, but I quickly saw that there were too many I'd like to see done with PDP8E's process.

One thing I wondered about was how often that site is updated. The most recent Spirit superres is from sol 1060, and the latest from Opportunity is sol 729. I know I've seen more recent superres sets come down from each of the rovers. It might be nice to see some of those.

Nice images--though I would suggest posting them in a more manageable file format, such as PNG or JPEG.