INFORGBEDGE_BASE - Base function for INFORGBEDGE

Contents

Syntax

   [map, mag, or, rmap, rmag, ror] = ...
             INFORGBEDGE_BASE(I, R, rho, sig, der, int);

See also

Ressembles: INFORGBEDGE, KOETHEDGE_BASE, CANNYEDGE_BASE. Requires: GSTSMOOTH_BASE, GRDSMOOTH_BASE, CANNYEDGEMAP_BASE.

Function implementation

function [map, mag, or, rmap, rmag, ror] = inforgbedge_base(I, R, rho, sig, der, int)

deal with multichannel image

[T,~,~,~,gy2,gxy] = ...
    gstsmooth_base(I, rho, sig, der, int, 1, [], false, false, [], []);

compute the eigen decomposition

[l1,l2,e1,e2] = gstdecomp(T);                                          %#ok
% D = sqrt(abs(gx2.^2 - 2*gx2.*gy2 + gy2.^2 + 4*gxy.^2));
% % compute edge orientation from first eigenvalue
% l1 = (gx2 + gy2 + D) / 2;
% % the 2nd eigenvalue would be:  l2 = (gx2 + gy2 - D) / 2;

compute the tensor norm

mag = sqrt(l1); % Koethe suggests sqrt(l1-l2);
mag = mag /max(mag(:));
% compute edge orientation (from eigenvector tangent)
or = atan2(-gxy, l1 - gy2);

feed it to a Canny-like algorithm

map = cannyedgemap_base(e1(:,:,1), e1(:,:,2), 'matlab', mag, or, [], []);

deal with reduced image

compute the single channel image gradient

[gx, gy] = grdsmooth_base(R, sig, der, [], 'ij');

feed it to a Canny-like algorithm

[rmap, rmag, ror] = cannyedgemap_base(gx, gy, 'matlab', [], [], [], []);

end % end of inforgbedge_base