Code covered by the BSD License

# Absolute Orientation - Horn's method

by

### Matt J (view profile)

23 Dec 2009 (Updated )

Solves weighted absolute orientation problem using Horn's quaternion-based method.

[regParams,Bfit,ErrorStats]=absorient(A,B,doScale)
```function [regParams,Bfit,ErrorStats]=absorient(A,B,doScale)
% This tool solves the absolute orientation problem, i.e., it finds the
% rotation, translation, and optionally also the scaling, that best maps one
% collection of 2D point coordinates to another in a least squares sense.
% Namely,
%
%            [regParams,Bfit,ErrorStats]=absorientParams(A,B,doScale)
%
% solves, when doScale=false (the default),
%
%                min. sum_i ||R*A(:,i) + t - B(:,i)||^2
%
% where R is a 2D rotation matrix and t is a translation vector.
%
%
%When doScale=true, it solves the more general problem
%
%                min. sum_i ||s*R*A(:,i) + t - B(:,i)||^2
%
%where s is a global scale factor. The registration uses Horn's
%quaternion-based algorithm.
%
%
%in:
%
%  A: a 2xN matrix whos columns are the 2D coords of N source points.
%  B: a 2xN matrix whos columns are the 2D coords of N target points.
%  doScale: Boolean flag. If true (default=false), the registration will
%          include a scale factor.
%
%out:
%
%
% regParams: structure output with estimated registration parameters,
%
%     regParams.R:   The estimated rotation
%     regParams.t:   The estimated translation
%     regParams.s:   The estimated scale factor (set to 1 if doScale=false).
%     regParams.M:   3x3 homogenous coordinate transform matrix [s*R,t;[0 0 1]].
%     regParams.q:   A unit quaternion [q0 qz] corresponding to R and
%                    signed so that R is a counter-clockwise rotation about
%                    the 3rd axis.
%
%
%  Bfit: The rotation, translation, and scaling (as applicable) of A that
%        best matches B in least squares sense.
%
%
% ErrorStats: structure output with error statistics. In particular,
%             defining err(i)=norm( Bfit(:,i)-B(:,i) ), it contains
%
%      ErrorStats.errlsq = 0.5* norm(err)
%      ErrorStats.errmax = max(err)
%
%
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Author: Matt Jacobson
% Copyright, Xoran Technologies, Inc.  http://www.xorantech.com

ncols=@(M) size(M,2); %number of columns
matmvec=@(M,v) bsxfun(@minus,M,v); %matrix-minus-vector

nn=ncols(A);

if nargin<3, doScale=false; end

if nn~=ncols(B),
error 'The number of points to be registered must be the same'
end

lc=mean(A,2);  rc=mean(B,2);  %Centroids

left  = matmvec(A,lc); %Center coordinates at centroids
right = matmvec(B,rc);

M=left*right.';

Nxx=M(1)+M(4); Nyx=M(3)-M(2);

N=[Nxx   Nyx;...
Nyx   -Nxx];

[V,D]=eig(N);

[trash,emax]=max(real(  diag(D)  )); emax=emax(1);

q=V(:,emax); %Gets eigenvector corresponding to maximum eigenvalue
q=real(q);   %Get rid of imaginary part caused by numerical error

q=q*sign(q(2)+(q(2)>=0)); %Sign ambiguity
q=q./norm(q);

R11=q(1)^2-q(2)^2;
R21=prod(q)*2;

R=[R11 -R21;R21 R11]; %map to orthogonal matrix

if doScale

summ = @(M) sum(M(:));

sss=summ( right.*(R*left))/summ(left.^2);
t=rc-R*(lc*sss);

else

sss=1;
t=rc-R*lc;

end

regParams.R=R;
regParams.t=t;
regParams.s=sss;
regParams.M=[sss*R,t;[0 0 1]];
regParams.q=q;

if nargout>1

Bfit=matmvec((sss*R)*A,-t);

end

if nargout>2

l2norm = @(M,dim) sqrt(sum(M.^2,dim));

err=l2norm(Bfit-B,1);

ErrorStats.errlsq=0.5*norm(err);
ErrorStats.errmax=max(err);

end

```