I¡¯m debating whether I should look at one star, using FWHM or Half Flux Diameter or a similar measure, or whether there is a way to have the software look at a larger subset of the whole image. ?If I use a Raspberry Pi I don¡¯t have a whole lot of processor power. ?It will be a good software challenge for me. ?Wavelet analysis might be faster than FFT and give good results. ?Wolfram and Mathematica are installed on the Pi. ?Jpeg file size is another possibility, as it might measure how much entropy is in the image.
Rick
toggle quoted message
Show quoted text
Rick,
? ?Maybe you could look for spatial frequencies, e.g. do a spatial Fourier transform and find a spatial frequency peak that corresponds to stars. That would be pretty processor intensive.
JPEG files are already encoded as Discrete Cosine Transform. You may be able to get spatial frequencies by examining the file. (Possibly helpful article here: <>) The JPEG will only give you spatial frequencies over 8x8 pixel blocks but that may be sufficient.
-Charlie Lasnier
