Why there is a difference between the dct computed by the MATLAB dct function and one computed by dct matrix multiplication?

The result is different, though both methods give inverse transform correctly.

Answer 1

[David Goodmanson]

Hi Mohsin,

x needs to be a column vector as you did, but you don't need to take the transpose of dctmtx:

x = 1:8

y1 = dct(x')
y2 = dctmtx(8)*x'
y1 - y2

x1 = idct(y1)
x2 = dctmtx(8)\y2

Same results, out and back.

[Mohsin Shah]

Thank you so much for correcting me. It worked.