I'm an EE graduate. My work is more CS than EE and all my colleagues so far come from a CS background.
Today we had a debate on the usefulness of frequency domain - one of my colleagues suggested that frequency domain is not necessary for DSP. I disagreed - I'm currently brushing up my skills on DSP and anyone would agree that Fourier transform is pretty much the second most fundamental thing you learn in DSP, after sampling theorem (in order of appearance ).
What is your take on it? Can one skip frequency domain and be a DSP guy?
DSP is a massive field and some parts will always require the frequency domain approach.
Im curious to hear what this guy bases his argument on. As far as i know operations like filtering are much faster in frequency domain (multiplication vs convolution)
Besides the technical implementations, fields like biomedical engineering requires frequency domain because some phenomenas can only be extracted from biological signals using the spectral representation.
Gosh, I was telling exactly that. I started "For example, if you take some signals and do this thing called convolution - in frequency domain it can be simply mult..." and the discussion ended because of lack of time.
Another example comes to my mind: patterns in an image. Complicated pattern-sniffing can often be replaced with observing it in frequency domain - Image Processing people can attest that I guess.