For source extraction for non-sidereally tracked observations where stars form streaks, the goal would be to generate a list of center points from detected stars that are compatible with astrometry.net's solve-field tool. In addition to stars, extracting scientifically interesting objects, such as points or streaks at angles different from the stars, is also important. Here is my proposed implementation for the first part, which I have mostly completed:
Gather more data (refer to issue #23). If the connectivity feature (issue #24) is implemented, it would be beneficial to have a list of connected streaks detailing lengths, angles, and thicknesses.
Employ statistics to identify and filter for similar slope angles and lengths.
Sort results by thickness.
Output a FITS xylist file for solve-field with center points of filtered streaks.
For scientifically significant observations, a second filtering set might be needed to detect circular blobs (tracked objects). As for streaks in different directions, filtering streak.streaks is straightforward. However, many interesting streaks will intersect other star streaks, suggesting a preliminary analysis of contour data (before analyzing streak slopes, etc.) to potentially identify multiple streaks within a contour.
I am keen to hear opinions and seek assistance, particularly regarding the second part (scientific data extraction). As far as I know, there is no easily accessible open-source library that performs source extraction in this manner. This could be the groundwork for a tool in this area.
Hello,
For source extraction for non-sidereally tracked observations where stars form streaks, the goal would be to generate a list of center points from detected stars that are compatible with astrometry.net's
solve-fieldtool. In addition to stars, extracting scientifically interesting objects, such as points or streaks at angles different from the stars, is also important. Here is my proposed implementation for the first part, which I have mostly completed:solve-fieldwith center points of filtered streaks.For scientifically significant observations, a second filtering set might be needed to detect circular blobs (tracked objects). As for streaks in different directions, filtering
streak.streaksis straightforward. However, many interesting streaks will intersect other star streaks, suggesting a preliminary analysis of contour data (before analyzing streak slopes, etc.) to potentially identify multiple streaks within a contour.I am keen to hear opinions and seek assistance, particularly regarding the second part (scientific data extraction). As far as I know, there is no easily accessible open-source library that performs source extraction in this manner. This could be the groundwork for a tool in this area.
Best regards, Michael