A CORDIC based array architecture for complex discrete wavelet transform

In this paper an array architecture for computation of Complex Discrete Wavelet Transform has been proposed. The wavelet filter coefficients are realized using multiplier less pipelined CORDIC algorithm. The choice of pipelined CORDIC algorithm over the conventional one for realizing the filter coefficient of CDWT is hardware effective and also effects in high frequency operation. The controller unit clusters input samples into even and odd samples coming in proper sequence at each clock cycles. This clustering provides a good amount of parallelism for faster operation of the filter compared to direct filter realization. The 8-tap filter bank is implemented using array architecture, effecting in high throughput. The algorithm developed is implemented on FPGA using the Virtex XCV100 series.

[1]  Julian Magarey,et al.  Motion estimation using a complex-valued wavelet transform , 1998, IEEE Trans. Signal Process..

[2]  Julian Magarey,et al.  An improved motion estimation algorithm using complex wavelets , 1996, Proceedings of 3rd IEEE International Conference on Image Processing.

[3]  Julian Magarey,et al.  Multiresolution stereo image matching using complex wavelets , 1998, Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170).

[4]  S. Kung,et al.  VLSI Array processors , 1985, IEEE ASSP Magazine.

[5]  I. Daubechies Ten Lectures on Wavelets , 1992 .

[6]  Stéphane Mallat,et al.  A Theory for Multiresolution Signal Decomposition: The Wavelet Representation , 1989, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Shen-Fu Hsiao,et al.  Householder CORDIC Algorithms , 1995, IEEE Trans. Computers.

[8]  He-Ping Pan,et al.  General stereo image matching using symmetric complex wavelets , 1996, Optics & Photonics.