function T = perform_voting(M,sigma,c,n,p,thresh)

% perform_voting - perform the voting process.
%
%   T = perform_voting(M,sigma,c,n,p,thresh);
%
%   'M' is a tensor field created by the user.
%
%   - The oriented features of the image must be stored
%     in the *highest* eigenvectors of the tensors.
%   - The confidence in these featres is reflected by
%     ratio of the eigenvalues 
%     (l1/l2==1 imply no confidence at all, l2=0 imply total confidence).
%
%   Optional:
%   'sigma' and 'c' control the stick field (see 'compute_stick_tf').
%   'n' is the size of the kernel (typically smaller than the size of the image).
%   'p' is the precision for the computation of the ball tensor field.
%   'thresh' is a threshold (perform voting only if l1-l2>thresh where (l1,l2) are the eigenvalues).
%
%   See also perform_voting_ball.
%
%   Copyright (c) 2004 Gabriel Peyr