Danillo Bracco Graziosi authored Dec 12, 2018 and David Flynn committed Feb 05, 2019

This replaces the previous floating point transform implementation with
a fixed-point alternative with essentially identical compression
performance.