M. Vishwanath et al., VLSI ARCHITECTURES FOR THE DISCRETE WAVELET TRANSFORM, IEEE transactions on circuits and systems. 2, Analog and digital signal processing, 42(5), 1995, pp. 305-316
A class of VLSI architectures based on linear systolic arrays, for com
puting the 1-D Discrete Wavelet Transform (DWT), is presented. The var
ious architectures of this class differ only in the design of their ro
uting networks, which could be systolic, semisystolic, or RAM-based. T
hese architectures compute the Recursive Pyramid Algorithm, which is a
reformulation of Mallat's pyramid algorithm for the DWT. The DWT is c
omputed in real time (running DWT), using just N-omega(J-1) cells of s
torage, where N-omega is the length of the filter and J is the number
of octaves. They are ideally suited for single-chip implementation due
to their practical I/O rate, small storage, and regularity. The N-poi
nt 1-D DWT is computed in 2N cycles. The period can be reduced to N cy
cles by using N-omega extra MAC's. Our architectures are shown to be o
ptimal in both computation time and in area. A utilization of 100% is
achieved for the linear array. Extensions of our architecture for comp
uting the M-band DWT are discussed. Also, two architectures for comput
ing the 2-D DWT (separable case) are discussed. One of these architect
ures, based on a combination of systolic and parallel filters, compute
s the N-2-point 2-D DWT, in real time, in N-2+N cycles, using 2NN(omeg
a) cells of storage.