In the introduction section of a thesis I read a little about classic orthonormal transforms such as Fourier, discrete cosine and wavelet transforms and their application in signal processing. Then the term "complete transforms" vs "overcomplete" ones is introduced and a comparison has been made. I'd like to know more about the topic and I'd like to read more sources (preferably not mathematical books that go into mathematical details) with the following information:
- It must cover signal processing context which is my research topic.
- Definition of complete and overcomplete transforms, their applications, differences and similarities
My main field of interest is online/active dictionary learning algorithms (e.g. KSVD, AKSVD, PAUDL, OSDL, etc.) but I am trying to get familiar with such mathematical concepts as a prerequisite.