VHDL Implementation of Multiplierless, High Performance DWT Filter Bank
M. M. Aswale, Ritu Patil · World Congress on Engineering · 2007
The JPEG 2000 image coding standard employs the biorthogonal 9/7 wavelet for lossy compression. The performance of hardware implementation of 9/7-filter bank depends on accuracy and efficiency. A method for processing signal in a canonical signed digit format, filter due to fractional coefficient includes the steps of obtaining real coefficients optimized to desired filter characteristics, calculating scaling factors for each real coefficients which minimizes errors between the real coefficients and converted canonical signed digit codes, producing optimum canonical signed digit coefficients by using the calculated scaling factors and filtering input data by using the optimum canonical signed digit form coefficient. For multiplierless operation canonical sign digit plays an important roll in processing signal. Advantage of this implementation is to reduce the number of adders in hardware and also increase the speed of operation. Filter bank filters the low resolution information using lowpass filter and high resolution information by using high pass filter. Wavelet co-efficient decides the characteristic of filter. In this method we used 9/7 wavelet transform to implement multiplierless operation. To implement multiplierless Hardware we used sum or difference of power two-representation i.e. canonical signed digit form. The canonical signed digit form is optimal in terms of how many nonzero digits are required to represent a given number. On an average, a canonical representation used two thirds as many nonzero digits as a binary representation, so the canonical signed digit form can significantly reduce the number of adders required in hardware.