Video filtering with Fermat number theoretic transforms using residue number system

We investigate image and video convolutions based on Fermat number transform (FNT) modulo q=2/sup M/+1 where M is an integer power of two. These transforms are found to be ideal for image convolutions, except that the choices for the word length, restricted by the transform modulus, are rather limited. We discuss two methods to overcome this limitation. First, we allow M to be an arbitrary integer. This gives much wider variety in possible moduli, at the cost of decreased transform length of 16 or 32 points for M<32. Nevertheless, the transform length appears still to be useful especially with block-based image and video filtering applications. We call these transforms the generalized FNT (GFNT). The second solution is to use a residue number system (RNS) to enlarge the effective modulus, while performing actual number theoretic transforms with smaller moduli. This approach appears to be particularly useful with moduli q/sub 1/=2/sup 16/+1 and q/sub 2/=2/sup 8/+1, which allow transforms up to 256 points with a dynamic range of about 24 bits. We design an efficient reconstruction circuit based on mixed radix conversion for converting the result from diminished-1 RNS into normal binary code. The circuit is implemented in VHDL and found to be very small in area. We also discuss the necessary steps in performing convolutions with the GFNT and evaluate the integrated circuit implementation cost for various elementary operations.

[1]  N. Ahmed,et al.  FAST TRANSFORMS, algorithms, analysis, applications , 1983, Proceedings of the IEEE.

[2]  Lars-Inge Alfredsson,et al.  VLSI Architectures and Arithmetic Operations with Application to the Fermat Number Transform , 1996 .

[3]  Amar Aggoun,et al.  Area-time efficient diminished-1 multiplier for Fermat number transform , 1994 .

[4]  L. Leibowitz.  A simplified binary arithmetic for the Fermat number transform , 1976 .

[5]  Charles M. Rader,et al.  Fast transforms: Algorithms, analyses, applications , 1984 .

[6]  M. Bhattacharya,et al.  Number theoretic techniques for computation of digital convolution , 1984 .

[7]  H. M. Shao,et al.  A parallel VLSI architecture for a digital filter of arbitrary length using Fermat number transforms , 1982 .

[8]  C.C. Lamb,et al.  A table-lookup scheme for residue-to-binary conversion , 1995, Conference Record of The Twenty-Ninth Asilomar Conference on Signals, Systems and Computers.

[9]  S. Andraos,et al.  A new efficient memoryless residue to binary converter , 1988 .

[10]  Chip-Hong Chang,et al.  New efficient residue-to-binary converters for 4-moduli set {2/sup n/ - 1, 2/sup n/, 2/sup n/ + 1, 2/sup n+1/ - 1} , 2003, Proceedings of the 2003 International Symposium on Circuits and Systems, 2003. ISCAS '03..

[11]  J. Pollard,et al.  The fast Fourier transform in a finite field , 1971 .

[12]  Gian Carlo Cardarilli,et al.  Efficient modulo extraction for CRT based residue to binary converters , 1997, Proceedings of 1997 IEEE International Symposium on Circuits and Systems. Circuits and Systems in the Information Age ISCAS '97.

[13]  Yuke Wang.  Residue-to-binary converters based on new Chinese remainder theorems , 2000 .

[14]  S. Ge,et al.  Adaptive control of uncertain Chua's circuits , 2000 .

[15]  A. B. Premkumar,et al.  A formal framework for conversion from binary to residue numbers , 2002 .

[16]  P.V. Ananda Mohan,et al.  Novel RNS to binary converters , 1991, 1991., IEEE International Sympoisum on Circuits and Systems.

[17]  Sebastiano Impedovo,et al.  Residue-to-binary conversion by the "quotient function" , 2003, IEEE Trans. Circuits Syst. II Express Briefs.

[18]  A. Hiasat,et al.  New memoryless, mod (2n±1) residue multiplier , 1992 .

[19]  Tuukka Toivonen,et al.  NUMBER THEORETIC TRANSFORM -BASED BLOCK MOTION ESTIMATION , 2002 .

[20]  A.G.J. Holt,et al.  Practical implementations of block-mode image filters using the Fermat number transform on a microprocessor-based system , 1988 .

[21]  R. Krishnan,et al.  Computation of complex number theoretic transforms using quadratic residue number systems , 1986, ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[22]  M. Soderstrand,et al.  An improved residue number system digital-to-analog converter , 1983 .

[23]  A. Bouridane,et al.  Diminished-1 multiplier for a fast convolver and correlator using the Fermat number transform , 1988 .

[24]  Olli Silven,et al.  A NEW ALGORITHM FOR FAST FULL SEARCH BLOCK MOTION ESTIMATION BASED ON NUMBER THEORETIC TRANSFORMS , 2002 .

[25]  Graham A. Jullien,et al.  Number Theoretic Techniques in Digital Signal Processing , 1991 .

[26]  Graham A. Jullien,et al.  An efficient tree architecture for modulo 2n+1 multiplication , 1996, J. VLSI Signal Process..

[27]  N. Burgess.  Efficient RNS to binary conversion using high-radix SRT division , 1998, Conference Record of Thirty-Second Asilomar Conference on Signals, Systems and Computers (Cat. No.98CH36284).

[28]  Yuke Wang,et al.  A study of the residue-to-binary converters for the three-moduli sets , 2003 .

[29]  Graham A. Jullien,et al.  Hardware implementation of convolution using number theoretic transforms , 1979, ICASSP.

[30]  Amitabha Das.  A novel efficient parallel algorithm for RNS to binary conversion for arbitrary moduli set , 1994, Proceedings of ICASSP '94. IEEE International Conference on Acoustics, Speech and Signal Processing.

[31]  Dimitrios Soudris,et al.  VLSI methodology for the design of RNS and QRNS full adder based converters , 2002 .

[32]  W. R. Moore,et al.  Improved mixed-radix conversion for residue number system architectures , 1991 .

[33]  Neil Burgess,et al.  Scaled and unscaled residue number system to binary conversion techniques using the core function , 1997, Proceedings 13th IEEE Sympsoium on Computer Arithmetic.

[34]  Daniel Minoli,et al.  Mersenne numbers rooted on 3 for number theoretic transforms , 1980, ICASSP.

[35]  B. Vinnakota,et al.  Selection of bases for a residue number system , 1994 .

[36]  R. Krishnan,et al.  Implementation of complex number theoretic transforms using quadratic residue number systems , 1986 .

[37]  M. Etzel,et al.  The design of specialized residue classes for efficient recursive digital filter realization , 1982 .

[38]  G. Jullien,et al.  An improved residue-to-binary converter , 2000 .

[39]  Chip-Hong Chang,et al.  Design of a high speed reverse converter for a new 4-moduli set residue number system , 2003, Proceedings of the 2003 International Symposium on Circuits and Systems, 2003. ISCAS '03..

[40]  V. Dimitrov,et al.  Two-Dimensional Transforms Using Number Theoretic Techniques , 1996 .

[41]  G A Jullien,et al.  Residue number system implementations of number theoretic transforms in complex residue rings , 1980 .

[42]  H. Nussbaumer.  Fast Fourier transform and convolution algorithms , 1981 .

[43]  Fred J. Taylor.  An RNS discrete Fourier transform implementation , 1990, IEEE Trans. Acoust. Speech Signal Process..

[44]  Benjamin A. Premkumar,et al.  Combinatorial logic based forward converters in residue number systems , 2000, 2000 IEEE International Symposium on Circuits and Systems. Emerging Technologies for the 21st Century. Proceedings (IEEE Cat No.00CH36353).

[45]  J. McClellan,et al.  Hardware realization of a Fermat number transform , 1976 .

[46]  A. Hiasat.  Efficient residue to binary converter , 2003 .

[47]  H. Toyoshima,et al.  Hardware implementation of Chinese remainder theorem using redundant binary representation , 1995, VLSI Signal Processing, VIII.

[48]  Wei Wang,et al.  An area-time-efficient residue-to-binary converter , 2000, Proceedings of the 43rd IEEE Midwest Symposium on Circuits and Systems (Cat.No.CH37144).

[49]  V. Dimitrov,et al.  Generalized Fermat-Mersenne number theoretic transform , 1994 .

[50]  Rudolph E. Thun,et al.  On residue number system decoding , 1986, IEEE Trans. Acoust. Speech Signal Process..

[51]  C. Burrus,et al.  Fast Convolution using fermat number transforms with applications to digital filtering , 1974 .