Just what is Resample used for?

[sreilly]

I guess this comes under the "newbie" column even though I've been imaging for years. I often see the term "re-sample" used and have no idea what this is used for. I just looked at an image that the information about it stated the data was re-sampled from 2.6" to 2.2". I'm thinking that this is a form of sharpening or star size reduction but haven't any real clue. The reason this comes up is the image I've been working on. I usually try to use the data that is under 2.5" to create my images. That, for this area is pretty restrictive in Central Virginia. The Bubble nebula that I posted reference to in my last few posts used this restriction and that resulted in about 30% of the luminance data to be rejected. The color data is even more so probably at 60%. I have been imaging all at 1x1 bin rate where in the past I use to bin the color 2x2.  My image scale is .48" per pixel using my ST10 and 12.5" RC at f/9. Out of 34 color images binned at 1x1, the range is from 2.22" to 3.38" with only 9 of the 34 (25%) being under 2.5". The luminance data was from 1.86" to 3.41 in 53 images with 44 of that being under 2.5" (80%).

The real question is am I throwing away usable data that could be used? In particular the color data which is, for me anyway, just adding the color component and not detail to the image. I know that any data that is taking in "worse" seeing will result in more bloated stars which is something I don't want. But if some of this data can be used with a processing method to reduce the "bloat" can it or should it be used?

[Juan Conejero]

Hi Steve,

the data was re-sampled from 2.6" to 2.2"

I don't know the particular image and instrumental used in this example, but in the risk of misunderstanding it, I think 'resampling' here means that the original (raw) image has been oversampled from a scale of 2".6 per pixel to a scale of 2".2 per pixel. The word 'sampling' here refers to an image processing technique (recalculation of image sample values on a different lattice by interpolation) instead of acquisition conditions (the image scale at the focal plane, or the ratio between the image scale and the resolution of the instrument).

The real question is am I throwing away usable data that could be used?

I am not the correct person to answer this (I am a development guy), but I'd say that as long as you have enough raw frames as to achieve a reasonably good SNR, then you are doing it well by discarding low-quality frames. I think it's a matter of achieving a good compromise between SNR and resolution for a particular set of raw data. Let's see what other users with more observational knowledge have to say about this.