An error minimizer will compute a transformation matrix such as to minimize the error between the reading and the reference. More...
#include <PointMatcher.h>
Classes | |
| struct | ErrorElements |
| A structure holding data ready for minimization. The data are "normalized", for instance there are no points with 0 weight, etc. More... | |
Public Member Functions | |
| ErrorMinimizer () | |
| Construct without parameter. | |
| virtual | ~ErrorMinimizer () |
| virtual destructor | |
| T | getPointUsedRatio () const |
| Return the ratio of how many points were used for error minimization. | |
| T | getWeightedPointUsedRatio () const |
| Return the ratio of how many points were used (with weight) for error minimization. | |
| virtual T | getOverlap () const |
| If not redefined by child class, return the ratio of how many points were used (with weight) for error minimization. | |
| virtual TransformationParameters | compute (const DataPoints &filteredReading, const DataPoints &filteredReference, const OutlierWeights &outlierWeights, const Matches &matches)=0 |
| Find the transformation that minimizes the error. | |
Protected Member Functions | |
| ErrorElements & | getMatchedPoints (const DataPoints &reading, const DataPoints &reference, const Matches &matches, const OutlierWeights &outlierWeights) |
| Helper function outputting pair of points from the reference and the reading based on the matching matrix. | |
Static Protected Member Functions | |
| static Matrix | crossProduct (const Matrix &A, const Matrix &B) |
| Helper funtion doing the cross product in 3D and a pseudo cross product in 2D. | |
Protected Attributes | |
| T | pointUsedRatio |
| the ratio of how many points were used for error minimization | |
| T | weightedPointUsedRatio |
| the ratio of how many points were used (with weight) for error minimization | |
| ErrorElements | lastErrorElements |
| memory of the last computed error | |
An error minimizer will compute a transformation matrix such as to minimize the error between the reading and the reference.
Typical error minimized are point-to-point and point-to-plane.
1.6.3