I've been learning about blind signal separation.
I have noticed that there are a number of algorithms for performing blind signal separation. (eg: Independent component analysis)
Which begs the question, are there any 'non-blind' signal separation algorithms?
I've been searching for a while and I haven't found anything. Nothing is listed on the wiki page and I've never noticed any in open libraries such as open CV.
I'm not even sure what search terms to look for.
The opposite of blind-signal separation is not "non-blind signal separation". The thing is, depending on what you know on the signals you will use different algorithms.
For example, if you know that there is a low frequency signal mixed with a higher frequency one, you can use low/high-pass filtering to get the two signals.
There's no real name for what you're looking for because every case is different. The generic way is more or less filtering the signals with the information you know. But the filtering is always different.
The key point of blind signal separation is the separation of a mixture of signals when there is mathematically insufficient knowledge to do so. On the contrary if you, mathematically, have sufficient information to distinguish two or more signals from a given mixture of them, then you can, as mathematically expected, perfectly separate them using well known fundamental operations of signal processing (algebra, calculus, etc.)
This latter operation is not coined a specific name for it, since its principles directly follow that of well known basic signal processing manipulations. Blind source separation however makes some claims about the type of mixture and imposes some structures on the characteristics of the signals, which is used to compensate the lack of information necessary in the separation of the mixture. That's why it is given a specific name.
Source separation can be attributed several adjectives. The reason is the structure of the sources, and the way they are mixed, linear or not, instantaneous or convolutive, etc. With convolutive mixtures, the blindness arises from the absence or very reduced level of knowledge on the sources and the convolution filters. Some used myopic, or informed source spearation, to qualify the degree of information one knows about either the sources or the filtering system.
They keywords you want to google are "supervised source separation" and "semi-supervised separation". Basically, if you have some information about the signals, you can typically leverage that to get better performance than blind source separation.
For example, let's say you know one signal in an audio recording is a trumpet. You could train a neural network on a dataset of trumpet sounds. Then feed your recording to the neural net and have it separate out the trumpet better than blind source separation algorithms. This would be an example of supervised source separation I believe.
