OT but, I sometimes see that with strong red/blue colours together as well (without colour blindness).

I believe that it's due to chromatic aberration of the lens in the eye. As the focus point of the red and blue is different, your brain interprets them as being at different distances.

James

I'm not sure that's a good idea - you do want Oppy to get to Endeavour don't you?

Seriously, thanks for all the nice comments.


Great work Fran, Mike, Geert, very impressive.

I'm glad we all come up with similar results - that's comforting.

Handy side effect of us all using the same bits of the jp2 I cut up - nice and easy to compare them.

James

Here is the analysis applied to the full (and lower ) resolution HiRISE image over the region just south of Victoria.

This shows what the algorithm does and doesn't detect, and the kind of terrain each colour represents.



James

OK Paolo here is what my Fourier transform analysis technique throws out.



Three versions- raw gray-scale, the 255 gray-scale levels turned into colour green->blue->red and a version trying to reduce it down to just three terrain types. No or very small ripples = Green, Ripples = Blue, Large Ripples of Purgatory size or greater = Red.

Yes, I know this is not exactly what Paolo wants, Red = exclusively curved Purgatory type ripples but I think that red areas both contain the majority of these and also are the areas where the ripples are so large that maneuvering around 'Purgatory types' would be difficult, and may well cause just as many problems as 'Purgatories' when trying to move across them in any direction but due South.

I think this gives a good overview, in particular picking out several 'rays' outward from Victoria with smaller ripples compared to the surroundings.

James

Now to convert those JP2's to JPG's for Juramike - if anyone else wants them, PM me.

Oh yes, I break the jp2 into chunks 5152 pixels high so you'll get it in those size chunks - but 29794 x 5152 is still a HUGE jpg.



I'm at that stage now and am assuming:

Green - no ripples or very (Insignificantly) small.
Blue - Ripples smaller than Purgatory (or maybe even the ripples around Purgatory if not enough comes out red)
Red - Larger than Blue ripples

I'll let you know when its done.

I'll see what I can do - it will be pretty large though.

Where would we be without your crazy ideas?

Keep 'em coming - with just a small dose reality we've ended up with your fantastic (and probably accurate ) last diagram.

Onwards!

J

This is a guess, I'm at work and don't have the time or the tools to do a proper analysis so the scale could be well off. But I would say it should be more like this...

Maybe, but I'm not sure quite how accurate the image will be. There are multiple THEMIS images which don't line up perfectly which will cause an error and I haven't reprojected the navcam image so that will be slightly off. I wouldn't rule out your original guess yet.




Yes Rui's modified Cook is right, that I am sure of. This analysis points to that hill, as does the different method I used on different images two years ago. To be clear my objection in post 2 was to Rui's original image which he subsequently updated.

Anyone remember those 'Inverse polar' images I did to match up the far rim of Victoria with the HiRISE images when we arrived?



Well I thought is was about time to dust off that technique.



Each column of the image represents one line of sight from Duck Bay across the THEMIS images - each being 0.044 degrees apart (the same as the navcam resolution) the left edge is at 45 degrees (North-East) and the right edge at 180 degrees (South)

At the top is a crop and 5x Phil-o-vision of the image Ustrax posted, According to the tracking site the centre of that image is at 108 degrees so that is where I've placed it - no fudging to make it line up required!

James

Nice, well spotted.

One of my images in there too.

It's been a long time - I may have to recheck it sometime, but last time I did this (over two years ago!) I reckoned that "Cook" was the hill further to the right of where you place it. (EDIT: now as I had it in the updated version)

Great names, great thread. This is going to be fun.

James

Now I've convinced myself that the Fourier method works it's time to look ahead.

Here is the 'difficult bit' Between Victoria and the green pastures beyond. Including a stupidly large full resolution version.



Looks like there are a few paths through.

James

Well I wasn't planning on making maps at the full HiRISE resolution, it is just too big. The 1/4 version is enough to see the ripples. I may do a few sections at higher resolution for the area just south of Victoria that we are likely to be crossing soon, but for general route planning I think this scale (1m/pixel) is better.

Of course I've posted my raw FT map at it's full resolution (8m/pixel) and you have the full resolution HiRISE, so if anyone wants to play with huge images then go ahead.



It's a matter of how you interpret the map. It's not necessarily red=bad, green=good. Red should be fine if you stick to the troughs, Oppy got through some of the worst areas already! It's more "you'll be limited in the direction of travel and maneuverability" rather than "bad" So 1) in red/orange areas Oppy will probably be limited to heading south 2) If she encounters any obstacles (large ripples, dead end trough, etc.) it will be harder to go around them.

I don't see the point of mapping every ripple - they are obvious in the raw HiRISE alone.

James

That's good to know Phil, thanks.

I've made up colour overlays at 1/4 of the full HiRISE resolution for the whole image so you can see how it's doing.

I'm quite frankly astonished at how good it is at picking out the ripple size. The resolution of this new version is now good enough that it can (and does) pick out the more isolated large ripples, I think even Purgatory is edging towards "red". I'm really pleased with how it's turned out, hopefully it will come in useful for the adventure ahead.



Enjoy. I could spend hours looking at them but I really must go to bed...

GO OPPY!

With the new method of only shifting 32 pixels each time a small error occurred in the version I posted earlier. One line of pixels was missing at each transition from one subimage to the next (I've broken the JP2 into 15 smaller JP2's for memory reasons) I've now fixed this:


Click for PNG (1.6Mb)

I was just about to email the latest raw output image to Paolo, but in the spirit of publishing raw images, it struck me that others might find it useful too. So I'll post it here:


Click for PNG (1.6Mb)

James

Yeah, I suppose that would be useful. I'm tempted to wait for the new HiRISE images to come in first - I have a feeling it could all get very complicated very quickly down that route - and until I'm convinced of the benefit, I'm not sure I want to go there!

In the meantime, I've rerun things with 64x64 input pixels but also taking John Spencers suggestion and not shifting the full 64 pixels each time, just 32 (not one, I'm not ready for 10's of Billions of FT's!) I've also changed the range from green to red to stop so much of the Etched Terrain being saturated red.



Click image as usual.

The more I look at this the more optimistic I am that Oppy will make good progress.

High resolution versions coming later.

James

E-W done, N-S easy, anything else would require a fair amount of work or a rotation of the image (OK for small areas, but v difficult for the whole image) however it would probably be better to do 2D Fourier transforms. I'm not sure what you expect to gain. The ripple orientation would be nice but apart from anomalous 'Purgatory' types (that I don't think it would pick up) they are all so close to N-S that it makes no difference.



Good idea, shouldn't be too hard to implement - it would however increase the number of Fourier transforms I do from ~2 million to ~30 Billion! So maybe not every pixel, but I suppose I could overlap the FT boxes a bit. I'm still not even sure how much smaller I could make the boxes; 64 pixels should still cover a few ripples so it might work fine. So may things to try and so little time.

I could also just use a different interpolation when blowing up the output pixels back to the full HiRISE size if people prefer the blended look. I quite like the blocks as you can see exactly what terrain was used to make that particular colour - for checking the accuracy it essential to plot it that way IMO.

James

I don't think that is worth doing in too much detail very far in advance.

We could map out a great route through the maze - then they'll want to nip across a couple of ripples to get to an interesting cobble and the whole route may change.

I'll try and make those 1/4 resolution plots for the whole of image this evening. Needed an early night last night so I just did a couple to show that it works.



Yes, I had no real idea how big the numbers would get as it worked across the image so a lot of those in the etched terrain ended up saturated red. I'll rerun it with different scaling before making the maps tonight.



Well at this stage that was never going to happen - my analysis explicitly looks for repeating E-W patterns at 8 - 28 pixel wavelengths - then plots the 'power' at that wavelength.

I think it's a reasonable assumption that big/funky dunes are more likely in 'big ripple areas'. In those areas it's also harder to navigate around obstacles and to navigate through in general. More elaborate schemes are probably possible but I'm not sure it's needed.

Purgatory was one isolated bump, not only that but it is surrounded by ripples that are small compared to some of the monsters found elsewhere, so the orange classification is what I would have expected (orange -> use caution, no blind drives like at Purgatory!). As I said before, this method with give you the areas that are generally more favourable for driving - but it doesn't mean Paolo can take his eyes off the road!

Any map that had the resolution to have every 'Purgatory' and dead end marked would end up being on a similar scale to the HiRISE image, so you might as well just use it as is!

My map should help to decide the general direction to minimise the chance of tripping over something nasty - then the HiRISE images themselves can be used to further refine the route - then the previous sols nav/pancams can be used to plot the actual drive. Simple!


James

Here are a couple of 1/4 resolution images (click for my website) with my coloured overlay.

Looks good. In fact, considering the crudeness of what I have done, I'm astonished how well it does.

Just north of Erebus - I assume Purgatory is in there somewhere.


Where we are now.


James

OK folks, version 1 of my Fourier technique is done for the whole of PSP_009141_1780. I'm still to make lots of high resolution overlays to see just how well it's doing but from first glance it looks pretty good. Here is the coloured version of the HiRISE 'browse' image.



Click for my website with 2048x5380 and 1024x2690 sized versions.

James

EDIT: BTW, This covers all the route so far and then to beyond "mini Endurance"

If you keep going down this route you might end up with the Fourier technique that I'm using.

Mine effectively boils down to the strength of repeating pattens across parts of the image. Yours is starting to sound like that too.

I don't think we need, or even want, a method that tries to classify every ripple and looks for dead ends, etc. What I am trying to produce is a map that says 'Over here the ripples are generally small - you could go around any rouge Purgatories without too much difficulty' and 'Over here are a load of monsters - stay away or you could spend weeks getting through the maze'

A Purgatory type ripple doesn't need to be 'red' if it's easy to go around.

After all, this map is to help get a rough idea of the route Oppy should take to get to Endeavour the fastest - not to find every little danger to avoid (That will be Paolo's job as we go along!)

James

P.S. That's great work Fran - my only worry is that the exposed bedrock is always 'green' I'll be interesting to see how this works over the larger variety of terrain seen on larger scales.