HoughPeaks function

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

HoughPeaks function

ariel.solomon

Hi,

I have copied and assimilated the houghpeaks function into octave.

The function:

function [r, c, hnew] = houghpeaks(h, numpeaks, threshold, nhood)

%HOUGHPEAKS Detect peaks in Hough transform.

%   [R, C, HNEW] = HOUGHPEAKS(H, NUMPEAKS, THRESHOLD, NHOOD) detects

%   peaks in the Hough transform matrix H.  NUMPEAKS specifies the

%   maximum number of peak locations to look for.  Values of H below

%   THRESHOLD will not be considered to be peaks.  NHOOD is a

%   two-element vector specifying the size of the suppression

%   neighborhood.  This is the neighborhood around each peak that is

%   set to zero after the peak is identified.  The elements of NHOOD

%   must be positive, odd integers.  R and C are the row and column

%   coordinates of the identified peaks.  HNEW is the Hough transform

%   with peak neighborhood suppressed.

%

%   If NHOOD is omitted, it defaults to the smallest odd values >=

%   size(H)/50.  If THRESHOLD is omitted, it defaults to

%   0.5*max(H(:)).  If NUMPEAKS is omitted, it defaults to 1.

 

%   Copyright 2002-2004 R. C. Gonzalez, R. E. Woods, & S. L. Eddins

%   Digital Image Processing Using MATLAB, Prentice-Hall, 2004

%   $Revision: 1.5 $  $Date: 2003/11/21 13:34:50 $

 

if nargin < 4

   nhood = size(h)/50;

   % Make sure the neighborhood size is odd.

   nhood = max(2*ceil(nhood/2) + 1, 1);

end

if nargin < 3

   threshold = 0.5 * max(h(:));

end

if nargin < 2

   numpeaks = 1;

end

 

done = false;

hnew = h; r = []; c = [];

while ~done

   [p, q] = find(hnew == max(hnew(:)));

   p = p(1); q = q(1);

   if hnew(p, q) >= threshold

      r(end + 1) = p; c(end + 1) = q;

 

      % Suppress this maximum and its close neighbors.

      p1 = p - (nhood(1) - 1)/2; p2 = p + (nhood(1) - 1)/2;

      q1 = q - (nhood(2) - 1)/2; q2 = q + (nhood(2) - 1)/2;

      [pp, qq] = ndgrid(p1:p2,q1:q2);

      pp = pp(:); qq = qq(:);

 

      % Throw away neighbor coordinates that are out of bounds in

      % the rho direction.

      badrho = find((pp < 1) | (pp > size(h, 1)));

      pp(badrho) = []; qq(badrho) = [];

 

      % For coordinates that are out of bounds in the theta

      % direction, we want to consider that H is antisymmetric

      % along the rho axis for theta = +/- 90 degrees.

      theta_too_low = find(qq < 1);

      qq(theta_too_low) = size(h, 2) + qq(theta_too_low);

      pp(theta_too_low) = size(h, 1) - pp(theta_too_low) + 1;

      theta_too_high = find(qq > size(h, 2));

      qq(theta_too_high) = qq(theta_too_high) - size(h, 2);

      pp(theta_too_high) = size(h, 1) - pp(theta_too_high) + 1;

 

      % Convert to linear indices to zero out all the values.

      hnew(sub2ind(size(hnew), pp, qq)) = 0;

 

      done = length(r) == numpeaks;

   else

      done = true;

   end

end

 

I get an error each time with different nhood values:

Input: P=houghpeaks(H,5,'threshhold',ceil(0.3*max(H(:))));

Error:

P=houghpeaks(H,5,'threshhold',ceil(0.3*max(H(:))));

error: nhood(2): out of bound 1

error: called from

    houghpeaks at line 44 column 10

 

Any thoughts?

Best,

Ariel

 



Reply | Threaded
Open this post in threaded view
|

Re: HoughPeaks function

Pantxo
ariel.solomon wrote

> Hi,
>
> I have copied and assimilated the houghpeaks function into octave.
>
> The function:
>
> function [r, c, hnew] = houghpeaks(h, numpeaks, threshold, nhood)
>
> %HOUGHPEAKS Detect peaks in Hough transform.
>
> %   [R, C, HNEW] = HOUGHPEAKS(H, NUMPEAKS, THRESHOLD, NHOOD) detects
>
> %   peaks in the Hough transform matrix H.  NUMPEAKS specifies the
>
> %   maximum number of peak locations to look for.  Values of H below
>
> %   THRESHOLD will not be considered to be peaks.  NHOOD is a
>
> %   two-element vector specifying the size of the suppression
>
> %   neighborhood.  This is the neighborhood around each peak that is
>
> %   set to zero after the peak is identified.  The elements of NHOOD
>
> %   must be positive, odd integers.  R and C are the row and column
>
> %   coordinates of the identified peaks.  HNEW is the Hough transform
>
> %   with peak neighborhood suppressed.
>
> %
>
> %   If NHOOD is omitted, it defaults to the smallest odd values >=
>
> %   size(H)/50.  If THRESHOLD is omitted, it defaults to
>
> %   0.5*max(H(:)).  If NUMPEAKS is omitted, it defaults to 1.
>
>  
>
> %   Copyright 2002-2004 R. C. Gonzalez, R. E. Woods, & S. L. Eddins
>
> %   Digital Image Processing Using MATLAB, Prentice-Hall, 2004
>
> %   $Revision: 1.5 $  $Date: 2003/11/21 13:34:50 $
>
>  
>
> if nargin < 4
>
>    nhood = size(h)/50;
>
>    % Make sure the neighborhood size is odd.
>
>    nhood = max(2*ceil(nhood/2) + 1, 1);
>
> end
>
> if nargin < 3
>
>    threshold = 0.5 * max(h(:));
>
> end
>
> if nargin < 2
>
>    numpeaks = 1;
>
> end
>
>  
>
> done = false;
>
> hnew = h; r = []; c = [];
>
> while ~done
>
>    [p, q] = find(hnew == max(hnew(:)));
>
>    p = p(1); q = q(1);
>
>    if hnew(p, q) >= threshold
>
>       r(end + 1) = p; c(end + 1) = q;
>
>  
>
>       % Suppress this maximum and its close neighbors.
>
>       p1 = p - (nhood(1) - 1)/2; p2 = p + (nhood(1) - 1)/2;
>
>       q1 = q - (nhood(2) - 1)/2; q2 = q + (nhood(2) - 1)/2;
>
>       [pp, qq] = ndgrid(p1:p2,q1:q2);
>
>       pp = pp(:); qq = qq(:);
>
>  
>
>       % Throw away neighbor coordinates that are out of bounds in
>
>       % the rho direction.
>
>       badrho = find((pp < 1) | (pp > size(h, 1)));
>
>       pp(badrho) = []; qq(badrho) = [];
>
>  
>
>       % For coordinates that are out of bounds in the theta
>
>       % direction, we want to consider that H is antisymmetric
>
>       % along the rho axis for theta = +/- 90 degrees.
>
>       theta_too_low = find(qq < 1);
>
>       qq(theta_too_low) = size(h, 2) + qq(theta_too_low);
>
>       pp(theta_too_low) = size(h, 1) - pp(theta_too_low) + 1;
>
>       theta_too_high = find(qq > size(h, 2));
>
>       qq(theta_too_high) = qq(theta_too_high) - size(h, 2);
>
>       pp(theta_too_high) = size(h, 1) - pp(theta_too_high) + 1;
>
>  
>
>       % Convert to linear indices to zero out all the values.
>
>       hnew(sub2ind(size(hnew), pp, qq)) = 0;
>
>  
>
>       done = length(r) == numpeaks;
>
>    else
>
>       done = true;
>
>    end
>
> end
>
>  
>
> I get an error each time with different nhood values:
>
> Input: P=houghpeaks(H,5,'threshhold',ceil(0.3*max(H(:))));
>
> Error:
>
> P=houghpeaks(H,5,'threshhold',ceil(0.3*max(H(:))));
>
> error: nhood(2): out of bound 1
>
> error: called from
>
>     houghpeaks at line 44 column 10
>
>  
>
> Any thoughts?
>
> Best,
>
> Ariel

Hi,

I think the error message tells you that you the NHOOD argument you have
supplied is scalar while the documentation states it should be a two-element
vector.

Pantxo



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-General-f1599825.html