I don't get the question. Lossy compression uses color quantization all the time, doesn't it?
Not typically. JPEG zeros out high frequency DCT coefficients, which is not the same thing. Hence my question: why go through this memory-intensive procedure?