Pathfinder site, for comparison with HIRISE Options

[Phil Stooke]

Presumably we will see a HIRISE image of the Pathfinder site at some point, so I am posting a polar view of the site here for comparison. I started with the Presidential Panorama. There was a slight mis-match between the ends of the pan - as if it had been slightly cropped for posting - which I can't fix, but it doesn't do too much damage to the polar view.

Phil

(PS - no, I can't really spare the time for this but you know how it is)

[djellison]

I'm still wondering how you did those - superb work, a data set that needed a proper 'going over' - and oh boy did you do that

Doug

[tedstryk]

I'm still wondering how you did those - superb work, a data set that needed a proper 'going over' - and oh boy did you do that

Doug

To tell it the short way, slowly.

To tell it the long way, the first part was figuring out how to put the puzzle together. The images were all transmitted in little subframe chunks, with the exception of four large images. Those four images all had areas that were horribly overexposed in some of the colors. Then I picked an image to use as my standard (after converting everything to 16 bit). I did my best to tweak the brightness and contrast of the other images until they matched. Also, in a few of the images, I had to correct for shadows moving between the framlets. In some of the very underexposed framelets, I had to do a lot of noise reduction, and I had to adjust the color to correct for the fact that there was an annoying "glow" in blue that skewed dark areas. Once this was done, I re-aligned the color channels. Most significantly, the green channel.

Doug, I think it was you who once asked about making red-blue images and chucking the green channel. This does not work well, because the CCD had far more green pixels than anything else, so the already really bad resolution gets worse. However, there was an annoying echo in the green channel, which had to be reconstructed for each image, and then subtracted. I greatly smoothed the color data, using photoshop's smart blur so as to blur together groups of pixels without bleeding occuring between the surface and pieces of the rover.

I wish I couldh have automated the process of cleaning and aligning the channels. However, nearly all the mosaics of framelets only cover a fraction of the CCD, and there is no information on the PDS cd to indicate where on the CCD each one is from. Since the severity and the direction of the distortion varies throughout the image, one had to manually assess the situation for each image.

One problem was the fact that the camera used some type of auto-stretching. So, if there was a glint off a piece of a rover in an image, the surrounding area would be nearly black. Then, in the next image, with no glint, it would be a nice, bright, well exposed image. With no guide to calibrate the rebalancing process, a lot of guesswork was required. Also, because the bright areas were totally saturated (most of the time), it required a lot of guesswork to reconstruct the greyscale to match other images. Often, pieces of the rover were washed out in some of the channels. That is why color varies in some of the images...a standard was really hard to find.

Soon, I hope to do the much easier task of working with the front camera images.