CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

can you provide the .cpp as attachement, nabble is eating some special 
characters.

Sebastien.

On 04/03/2012 06:52 PM, edward wrote:
> Hi everyone,
> The function linear_least_squares_fitting_3 to estimate best fitting plane
> seems to obtain unexpected results when the points are generated from a
> specific plane.
> I randomly constructed a plane and then retrived some points from this
> plane. Another new plane is estiamted by these points utilizing funtion
> linear_least_squares_fitting_3. But these two planes are not the same!
> Here are main codes:
>
>   boost::mt19937 eng;  // a core engine class：Mersenne
>         eng.seed(time(NULL));
>         boost::uniform_real<>   unif01(-1,1); //<>不可少
>         boost::uniform_real<>   unif100(-100,100);
>         boost::variate_generator<boost::mt19937&amp;, boost::uniform_real&lt;>   
> >
> random01(eng, unif01);
>         boost::variate_generator<boost::mt19937&amp;, boost::uniform_real&lt;>   
> >
> random100(eng, unif100);
>         Plane_3 plane(random01(),random01(),random01(),random100());
>         cout<<"Origin plane"<<plane&lt;&lt;endl;
>         list&lt;Point_3>  pts;
>         int N=10;
>         for(int i=0;i<N;++i)
>         {
>                 double x, y,z;
>                 x=random100();
>                 y=random100();
>                 xy2z(plane,x,y,z);//User defined function, from x or y to 
> retrive z from
> plane
>                 pts.push_back(Point_3(x,y,z));
>         }
>         for(list&lt;Point_3>::iterator 
> iter=pts.begin();iter!=pts.end();++iter)
>         {
>                 cout<<"Point to Plane Dist"<<CGAL::squared_distance(*iter,
> plane)&lt;&lt;endl;
>         }
>
>         Plane_3 plane1; 
>         cout&lt;&lt;endl;
>         cout&lt;&lt;&quot;CGAL Fitting result&quot;&lt;&lt;
> CGAL::linear_least_squares_fitting_3(pts.begin(),pts.end(),plane1,
> CGAL::Dimension_tag&lt;0>());
>         cout<<endl;
>         cout&lt;&lt;&quot;cgal plane&quot;&lt;&lt;plane1&lt;&lt;endl;
>         for(list&lt;Point_3>::iterator 
> iter=pts.begin();iter!=pts.end();++iter)
>         {
>                 //cout<<*iter<<endl;
>                 cout&lt;&lt;&quot;Point to Plane Dist
> &quot;&lt;&lt;CGAL::squared_distance(*iter, plane1)&lt;&lt;endl;
>         }
> The output is:
> Origin plane&lt;b>-0.456603 0.437936 -0.276328 -49.6504*
> Point to Plane Dist 0
> Point to Plane Dist 0
> Point to Plane Dist 4.23699e-028
> Point to Plane Dist 0
> Point to Plane Dist 0
> Point to Plane Dist 0
> Point to Plane Dist 1.65507e-030
> Point to Plane Dist 1.6948e-027
> Point to Plane Dist 0
> Point to Plane Dist 0
>
> CGAL Fitting result 1
> cgal plane *-0.661375 0.634336 -0.400252 -71.917*
> Point to Plane Dist 1.01102e-017
> Point to Plane Dist 2.73393e-018
> Point to Plane Dist 9.74678e-018
> Point to Plane Dist 1.0964e-015
> Point to Plane Dist 6.0825e-016
> Point to Plane Dist 1.66278e-016
> Point to Plane Dist 1.23536e-015
> Point to Plane Dist 2.27407e-016
> Point to Plane Dist 2.63125e-018
> Point to Plane Dist 3.43804e-016
> Press any key to continue . . .
>
> It seems that the fitting results, point to plane squared distance is good,
> but plane equation are quite different. How does this happens?
>
> Hope you could help
>
> Best regards
>
> Edward
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Question-about-linear-least-squares-fitting-3-tp4529464p4529464.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.