PCFE/PPM Center Question

[1ponders] (Paul) · #1 21-08-2006, 10:53 AM

When using either programs, what settings need to be changed when trying to process bmps of Uranus or Neptune. I'm having difficulty getting PCFE to lock on and keep Uranus central to the crop area.

Hitchhiker · #2 21-08-2006, 11:14 AM

Hi Paul, for small objects like Uranus you are going to have to reduce the value of Min Pixels. Min Pixels is the minimum number of pixels that pass the threshhold (ppmcentre default is 40) for the file to be processed. The ppmcentre (and PCFE) default Min Pixels value is 250. I'm guessing that Uranus is smaller than 250 pixels in your image.

Here's a quote from iceman re using ppmcentre/PCFE on Uranus

Quote:

I've also used -minpixels=80 for a small target (like Uranus).

iceman (Mike) · #3 21-08-2006, 11:19 AM

Yep, minpixels=80 will be fine for Uranus. If the image is very noisy because of high gain, then you'll have to change threshhold as well.

[1ponders] (Paul) · #4 21-08-2006, 11:58 AM

That's what I thought, and I reduced the Minpixels to 40, but didn't change the threashold. The bmps were pretty grainy, so should the threshhold go up or down?

Hitchhiker · #5 21-08-2006, 12:06 PM

Try lowering the threshhold. I had to do this on some noisy raw images of Saturn I had. It requires a bit of experimenting to get right.

The Threshhold tells ppmcentre how to identify the object of interest in your image. If you have noisy images the pixel value of the object is not as far above the noise level as a 'clean' image.

iceman (Mike) · #6 21-08-2006, 12:06 PM

up, to something like 80.

iceman (Mike) · #7 21-08-2006, 12:07 PM

hmm, conflicting with adam! I agree with Adam, that is, experiment

[1ponders] (Paul) · #8 21-08-2006, 12:07 PM

Hitchhiker · #9 21-08-2006, 12:47 PM

I'm not sure which is the right way to go, now!

Anyway, a little experimenting will fix it.

iceman (Mike) · #10 21-08-2006, 12:50 PM

I thought it would be up, because the level of noise is higher, so what is considered "background" (ie: threshhold) also needs to be set higher.

[1ponders] (Paul) · #11 21-08-2006, 02:32 PM

I hate being a guinea pig

Thanks guys, I'll experiment and see how I go.

bird (Anthony Wesley) · #12 23-08-2006, 07:42 PM

You can also try loading one frame from your video into an image editing program like The Gimp (or photoshop, I assume) and looking at the pixel values for your background and foreground objects. That can be a useful way to see what ppmcentre is going to do.

cheers, Bird

[1ponders] (Paul) · #13 25-08-2006, 05:16 PM

What do you mean, bird? Sorry but I am not the sharpest picket in the fence when it comes to some of this stuff

. I understand about reading the pixel values in photoshop, but how would I use this information to adjust the values in ppmcentre?

bird (Anthony Wesley) · #14 28-08-2006, 07:28 PM

oops, sorry for the late reply, I forgot all about this thread, sorry :-)

Each pixel in your image has an intensity value somewhere between 0 and 255 (0 = black and 255 = white). ppmcentre sets a threshhold to distinguish between "background" pixels and "planet" pixels, any pixel that's darker than the threshhold is a background pixel, and any pixel thats lighter then the threshhold is considered a "planet" pixel.

ppmcentre finds the centre of your image by averaging together all the "planet" pixels and finding the centre.

You can change the threshhold that ppmcentre uses by adding the "-threshhold=X" commandline parameter when you run ppmcentre, and set X to the threshhold value that you want to use.

Now, to see the values in your own data you have to load a typical frame into an image editing program (like the GIMP, or photoshop), convert the image to greyscale and then move the mouse over the pixels and see what values they have.

Then you can decide for yourself whether the default value that ppmcentre uses (30) is good enough.

cheers, Bird

[1ponders] (Paul) · #15 28-08-2006, 09:15 PM

Thanks for that bird. I'll give it a whirl