A new milestone has been reached! My software is now able to automatically identify the current satellite, even before the signal is good enough for a receiver to lock any transponder.
If you read my previous posts, you will know that I am identifying the satellite through a transponder pattern recognition against known satellite transponder lists, stored in a folder.
This satellite transponder lists can be generated and downloaded from different websites, like http://satellites-xml.eu and are included in the transponder_lists folder.
The results are impressive and actually better than I expected: the software identified all satellites I tried without any error and without any learning process. All I did was to download the current satellite transponder lists.
How does it work? Without going into deep, the software first determines the frequencies of the transponders in the live spectrum. Then it tries to find the satellite list, which contains the most matches to the identified transponders of the live spectrum. The satellite list with the most matches wins.
The percentage next to the satellite name shows the score: 100% would mean that all transponders in the live spectrum have been found in the winning satellite list. Mind you, a score of 50% is pretty good and will accuratly predict most satellites.
Here a few example pictures:
The identification score can be improved if the treshold is fine tuned: it represents the horizontal line, where transponders are identified. However, 60% is a pretty good default value!
Notice that the satellite identification tells you what band you are tuned to: VL, HL, VH or HH.
The name of the satellite takes about 1-2 seconds to be displayed and it just isn't quicker due to the fact that it takes about 3 seconds to refresh the live spectrum with 500 samples.
Interested in trying it out? What are you waiting for - download the latest release here: