 
  
  
  
  
Circular harmonic algorithms can be derived from the inversion formula of Cormack (1963). For the 2D problem Rf = g it is obtained by Fourier expansions
  
 
One can show that
  is the Chebysheff polynomial of the first kind of order
  is the Chebysheff polynomial of the first kind of order   .
In principle this defines an inversion formula for R: The Fourier coefficients
 .
In principle this defines an inversion formula for R: The Fourier coefficients   of 
g = Rf determine the Fourier coefficients
  of 
g = Rf determine the Fourier coefficients   of f via (5.1), hence f
is determined by g.
  of f via (5.1), hence f
is determined by g.
The formula (5.1) is useless for practical calculations since   increases exponentially with
  increases exponentially with   outside [-1,+1]. Cormack (1964) derived also a stable version of his inversion formula. It reads
  outside [-1,+1]. Cormack (1964) derived also a stable version of his inversion formula. It reads
  
 
where   is the Chebysheff polynomial of the second kind of order
  is the Chebysheff polynomial of the second kind of order   . This formula does not suffer from exponential increase. It is the starting point of the circular harmonic reconstruction algorithms of Hansen (1981), Hawkins and Barrett (1986), Chapman and Cary (1986).
 . This formula does not suffer from exponential increase. It is the starting point of the circular harmonic reconstruction algorithms of Hansen (1981), Hawkins and Barrett (1986), Chapman and Cary (1986).
We take a different route and start out from (2.1) again. We consider only the 
case n = 2. Putting   ,
 ,   , in (2.4) we obtain
 , in (2.4) we obtain
  
 
The j sum is a convolution. In order to make this convolution cyclic we extend   by putting
  by putting   , 
in accordance with the eveness property of the Radon transform. Then,
 , 
in accordance with the eveness property of the Radon transform. Then,
  
 
Defining
  
 
we have
  
 
This defines the circular harmonic algorithm.
Algorithm 4 (Circular harmonic algorithm for standard parallel geometry.)
 ,
 ,   ,
 ,
                 ,
 ,
g the 2D Radon transform of f.
 and extend
 
               and extend   to all
  to all   by
  
               by   .
 .
 ,
 ,   carry out
               the discrete cyclic convolutions
  carry out
               the discrete cyclic convolutions
                 
 
 and
  and   compute
  compute
                 
 
 is an approximation to
  is an approximation to   .
 .
The second step of the algorithm has to be done with a fast Fourier transform (FFT) 
in order to make the algorithm competitive with filtered backprojection. In that case step 2 requires O  operations. This is slightly more (by the factor
  operations. This is slightly more (by the factor   ) than what is needed in the filtered backprojection algorithm. The third step needs
 ) than what is needed in the filtered backprojection algorithm. The third step needs   additions.
  additions.
Algorithm 4 can be used almost without any changes for the interlaced parallel geometry,
i.e. for   with
  with   odd missing (p even). One simply puts
  odd missing (p even). One simply puts   for
  for   odd and doubles
  odd and doubles   in step 2.
  in step 2.
Circular harmonic algorithms are also available for standard fan beam data. (2.9) with   ,
 ,   ,
 ,   reads
  reads
  
 
where g is now the fan beam data function from (2.8).
Algorithm 5 (Circular harmonic algorithm for standard fan beam geometry.)
 ,
 ,   ,
 ,
                 with g from (2.8).
  with g from (2.8).
 .
 .  
 ,
 ,   carry out
               the discrete cyclic convolutions
  carry out
               the discrete cyclic convolutions
                 
 
 and
  and   compute the sums
  compute the sums
                 
 
 is an approximation to
  is an approximation to   .
 .
The complexity of Algorithm 5 is again O  .
 .
A few remarks are in order.
 ) than filtered backprojection they usually run 
          faster due to their simplicity. This is true in particular for 
          fan beam data because in that case the backprojection is quite time consuming.
 ) than filtered backprojection they usually run 
          faster due to their simplicity. This is true in particular for 
          fan beam data because in that case the backprojection is quite time consuming.
 
  
  
 