I How to determine an asteroid survey efficiency?

[pioneerboy]

I read in a couple of papers about asteroid hunting that they used artificially implemented asteroids in real images to see if their detection algorithms could find them. This way, the scientists determined their asteroid survey efficiency in real data.
As I don't have the resources to work so extensively with images, implement artificial stuff, and have a dedicated survey algorithm, I would like to know if there's another accurate way to determine the asteroid survey efficiency? You may well ask, WHAT FROM would I like to determine its efficiency if not from a survey algorithm. Well, that's the second question: using a telescope, CCD, typical calibration files to reduce noise, known celestial coordinates, time, etc. , a survey software like "Astrometrica", and a square-degree worth of images, how can I characterise the survey efficiency? After all, Astrometrica only shows already known objects, and has the ability to blink-compare images and mark presumably new objects. But it doesn't search for objects by itself so that a search efficiency could be characterised. What would be ways to characterise the search efficiency in a personal asteroid survey?

 

[mfb]

Can you make your software forget about objects? That way you can see if the software would have marked them as possible new object, and mimic the proper way to determine efficiency.

 

[pioneerboy]

What it does is to look up the orbital elements/epoch ephemerides of all objects that were in/around the image you took; so you basically know where to look for known ones without blink-comparing. What you have to do now is to investigate each object and select the appropriate object candidate from a whole list which is sorted according to RA/Dec distance from the marked object in the image; so, you'd want to go with the closest candidate (usually 0.1 to 0.2 away in both, RA/Dec). This is written into a .txt-file that is later sent to the MinorPlanetCenter for orbit improvement.
I'm not aware of any offline-detection-algorithm that is inside the software. So comparing offline vs. online ephemerides doesn't work I guess.
When hunting for new objects, all you really could do is visual blink-comparing. So in the end you'd never know how many new ones you detected from a total of present new objects in the image.

 

[pioneerboy]

Btw, an efficiency characterisation as in Fig.2 http://web.gps.caltech.edu/~mbrown/papers/ps/kbodeep.pdf would be great. If only there were a way around this artificial stuff and a more "amateur astronomical" approach without large amounts of GB Rams or CPUs.

 

[petex]

http://arxiv.org/abs/1511.07659