VLSI Implementation of Mallat's Fast Discrete Wavelet

Abstract

This paper proposes a novel VLSI architecture to compute the DWT (discrete wavelet transform) coefficients using Mallat's algorithm with reduced complexity. We studied the commonality embedded in the mirror filters of the algorithm and use a PLA as an Address Generator (PAG) to load the data for cascaded FIR computation. By using an embedded downsampling process in the control signal design, we reduced the complexity by saving storage and computation. The prototyping design is implemented and fabricated using AMI 1.5 micron CMOS process through MOSIS.