CANNYEDGEMAP - Find edge in an image from its directional derivatives.

Contents

Description

This function finds edges in a image, given its directional derivatives, and possibly the magnitude/orientation images. It is nothing else than Canny's approach [Canny86] (non-maxima suppression + hystheresis thresholding) applied on already estimated derivatives.

Algorithm

The Canny method finds edges by looking for local maxima of mag given as input. The method uses two thresholds, to detect strong and weak edges, and includes the weak edges in the output only if they are connected to strong edges. This method is therefore less likely than the others to be "fooled" by noise, and more likely to detect true weak edges.

Syntax

   map = CANNYEDGEMAP(gx,gy);
   map = CANNYEDGEMAP(gx,gy,'propertyname',propertyvalue,...)

Inputs

gx, gy : directional derivatives of an image in X- (horizontal) and Y- (vertical) directions, typically estimated by differentiating the Gaussian filter of an image; see GRDSMOOTH for discussion regarding the direction/orientation of the gradient derivatives.

Property [propertyname propertyvalues]

'mag' : a matrix storing typically the image gradient magnitude or the norm of the gradient structure tensor; if not parsed, it is the norm of the [gx,gy] gradient field (and possibly taking the max over the different channels if the gradient is 3D: see CANNYEDGES).

'or' : image containing feature normal orientation angles in degrees (0-180), angles positive anti-clockwise.

'der' : string setting the name of the function used for Canny edge detection; it can be:

default: der='edge'.

'hyst' : vector [low high] setting the low and high hystheresis threshold values with 0<=low<high<1 used with der='edges'; default: hyst=[], and the threshold values are chosen automatically (see EDGE) or using the variable ratio below.

'ratio' : vector with pixel and threshold ratios used for automatically selecting the hysteresis thresholds when those are not passed; default: ratio = [pixratio threshratio] = [1/3 0.08].

Output

map : binary edge map with non maximally suppressed image.

Acknowledgment

This function is a copy/paste and crop of several proposed functions implemented for thresholding, non maximum suppression and thinning of edge maps.

References

[Canny86] J.F. Canny: "A computational approach to edge detection", IEEE Trans. on Pattern Analysis and Machine Intelligence, 8(6):679-698, 1986. http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4767851&tag=1

[Kovesi] P.D. Kovesi: "MATLAB and Octave Functions for Computer Vision and Image Processing", The University of Western Australia. http://www.csse.uwa.edu.au/~pk/research/matlabfns/

See also

Ressembles: CANNYEDGE, EDGECORNER, CANNYEDGES, Requires: CANNYEDGEMAP_BASE.

Function implementation

function map = cannyedgemap(gx,gy,varargin)

parsing and checking parameters

error(nargchk(1, 20, nargin, 'struct'));
error(nargoutchk(1, 1, nargout, 'struct'));

if ~(isnumeric(gx) && isnumeric(gy))
    error('cannyedgemap:inputparameter','matrices required in input');
elseif ~isequal(size(gx),size(gy))
    error('cannyedgemap:inputparameter', ...
        'input matrices with same size required in input');
end

p = createParser('CANNYEDGEMAP');   % create an instance of the inputParser class.
% function handling the function used for estimating Canny edges
p.addParamValue('mag', [],  @(x)isfloat(x));
p.addParamValue('or', [], @(x)isfloat(x));
p.addParamValue('der', 'matlab', @(x)ischar(x) && ...
    any(strcmpi(x,{'vista','edge','matlab','kovesi'})));
p.addParamValue('hyst', [], ...
    @(x)isvector(x) && length(x)<=2 && all(x>=0) && all(x<1));
p.addParamValue('ratio', [1/3 0.08], ...
    @(x)isvector(x) && length(x)==2 && all(x>=0) && all(x<1));

% parse and validate all input arguments
p.parse(varargin{:});
p = getvarParser(p);

main computation

map = cannyedgemap_base(gx, gy, p.der, p.mag, p.or, p.hyst, p.ratio);

display

if p.disp
    figure, imagesc(edgemap), axis image off, title('edge map');
    if size(edgemap,3) == 1, colormap 'gray'; end;
end

end  % end of cannyedgemap