I'm currently working on a jpeg decoder based on the algorithm of Arai, Agui and Nakajima. While my solution is almost working I am missing a critical scaling operation that is key for this approach (which turns dct values into a dft values). I am supposed to scale the quantization matrix with based on the position of the values in the quantization matrix/ I just can't figure out how to calculate these scaling factors.
A good reference I am working with can be found here: http://vsr.informatik.tu-chemnitz.de/~jan/MPEG/HTML/IDCT.html
Hoping to find some people that have practical experience with this.
Thanks in advance!