Simplified biorthogonal discrete wavelet transform for VLSI architecture design

Abstract

Biorthogonal discrete wavelet transform (BDWT) has gained general acceptance as an image processing tool. For example, the JPEG2000 standard is completely based on the BDWT. In BDWT, the scaling (low-pass) and wavelet (high-pass) filters are symmetric and linear phase. In this work we show that by using a specific sign modulator the BDWT filter bank can be realized by only two biorthogonal filters. The analysis and synthesis parts use the same scaling and wavelet filters, which simplifies especially VLSI designs of the biorthogonal DWT/IDWT transceiver units. Utilizing the symmetry of the scaling and the wavelet filters we introduce a fast convolution algorithm for implementation of the filter modules. In multiplexer-demultiplexer VLSI applications both functions can be constructed via two running BDWT filters and the sign modulator.

Design of Orthogonal UWB Pulse Waveform for Wireless Multi-Sensor Applications

“…In continuous-time domain this corresponds to the modulation of the signal by cos( ) t  . The sign modulator has been used in VLSI wavelet transform chips to simplify the construction of the analysis and synthesis filters [4]. In z transform domain this equals to the replacement of the z variable by -z.…”