Hardware signal processing unit for one-dimensional variable-length discrete wavelet transform

Discrete wavelet transform is a powerful mathematical tool used for signal and image compression, nonlinear filtering or noise reduction, signal and biomedical image processing and all kind of applications that implies a time-located variation on signals. The development of a hardware signal processing unit for the discrete wavelet transform for computing one-dimensional variable-length input data signals, based on a filter bank algorithm, is shown in this paper. The system allows the selection of the mother-wavelet and depth level analysis by the user for a specific application. In order to keep open general applications of the system, a Xilinx Spartan-3 platform is used for implementation, based on its reconfigurability properties. The overall system performance is shown and it is compared with already available processing systems, applied to the same process. The novelty of this paper is to develop a general purpose, low-cost discrete wavelet transform core for embedded applications.