Things are proceeding more smoothly now; the locks on stars are holding (usually) through the entire 30-minute span of each dataset. There are some problems, but I came up with a nice ad-hoc solution: a new method called “updateStarPosition”. This permits me to provide the program with the correct positions of stars at any time that I think necessary. Of course, I have to go to the trouble of figuring out those positions, which can be a tedious process, but this turns out to be faster than endless fine-tuning of the lock algorithms. I have a good test for the solidity of the locks: at the end of each dataset, the program saves its positions for stars. I can then compare the program’s positions with the positions on an actual frame, looking for mismatches. In some cases, every single star is locked on properly, but in a few cases one or two stars are off the mark -- but seldom by much.
You might be asking, “Why is it so important to maintain a good star lock?” The answer is a bit roundabout. The basic measure of position in the frame is a simple rectangular Cartesian grid. The position of stars in the sky, however, is measured in spherical coordinates called right ascension and declination. If I know the right ascension and declination of a number of stars on the frame, as well as their x,y coordinates in the frame, I can deduce, by means of rather ugly calculations, the orientation of the spherical system for right ascension and declination in the frame. This in turn permits me to calculate the altitude and azimuth system for the frame. Finally, I can calculate the coordinate system whose pole lies on the radiant of the Leonid meteor shower. This is important because every Leonid travels along the vertical coordinates of this latter spherical coordinate system, enabling me to predict exactly where a Leonid should appear in future frames. I use this information to differentiate random collections of flares from actual Leonids. This is a very sensitive method; it can pick up faint Leonids that most people would miss. I wish I could illustrate this process with some diagrams -- that would make it much easier to understand. Unfortunately, diligent searching on the Internet was insufficient to find an appropriate image of a spherical coordinate system; the ones I could find were all too small for this purpose.
I have bitterly complained about The Idiot who keeps messing with the drape and ruining the images. Here’s an example of what I mean:
Images like this one are scattered through many datasets for camera AR50F; they have made life very difficult for me.
One last thing: I trace the edges of the field of view both to locate its precise center and to block out some Leonids that are so far off edge that their luminance measurements are utterly unreliable. Here’s a tracing of that edge: