boofcv.alg.geo.NormalizationPoint2D Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of boofcv-geo Show documentation
Show all versions of boofcv-geo Show documentation
BoofCV is an open source Java library for real-time computer vision and robotics applications.
/*
* Copyright (c) 2021, Peter Abeles. All Rights Reserved.
*
* This file is part of BoofCV (http://boofcv.org).
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package boofcv.alg.geo;
import georegression.geometry.UtilCurves_F64;
import georegression.struct.curve.ConicGeneral_F64;
import georegression.struct.point.Point2D_F64;
import georegression.struct.point.Point3D_F64;
import org.ejml.data.DMatrix3x3;
import org.ejml.data.DMatrixRMaj;
import org.jetbrains.annotations.Nullable;
/**
* Describes how to normalize a set of points such that they have zero mean and variance. This is equivalent
* to applying the matrix below. Normalization is often needed as a preprocessing step for solving linear equations.
* Greatly reduces bias and numerical errors.
*
*
* N = [ 1/σ_x 0 -μ_x/σ_x ]
* [ 0 1/σ_y 0 -μ_y/σ_y ]
* [ 0 0 1 ]
*
*
*
* Y. Ma, S. Soatto, J. Kosecka, and S. S. Sastry, "An Invitation to 3-D Vision" Springer-Verlad, 2004
*
*
* @author Peter Abeles
*/
public class NormalizationPoint2D {
// default value is do nothing
public double meanX = 0, stdX = 1;
public double meanY = 0, stdY = 1;
// internal workspace
private DMatrix3x3 work = new DMatrix3x3();
public NormalizationPoint2D() {}
public NormalizationPoint2D( double meanX, double stdX, double meanY, double stdY ) {
this.meanX = meanX;
this.stdX = stdX;
this.meanY = meanY;
this.stdY = stdY;
}
public void set( double meanX, double stdX, double meanY, double stdY ) {
this.meanX = meanX;
this.stdX = stdX;
this.meanY = meanY;
this.stdY = stdY;
}
/**
* Applies normalization to a H=3xN matrix
*
* out = Norm*H
*
* @param H 3xN matrix. Can be same as input matrix
*/
public void apply( DMatrixRMaj H, DMatrixRMaj output ) {
output.reshape(3, H.numCols);
int stride = H.numCols;
for (int col = 0; col < H.numCols; col++) {
// This column in H
double h1 = H.data[col], h2 = H.data[col + stride], h3 = H.data[col + 2*stride];
output.data[col] = h1/stdX - meanX*h3/stdX;
output.data[col + stride] = h2/stdY - meanY*h3/stdY;
output.data[col + 2*stride] = h3;
}
}
/**
* Applies normalization to a H=3xN matrix
*
* out = Norm*H
*
* @param H 3xN matrix. Can be same as input matrix
*/
public void remove( DMatrixRMaj H, DMatrixRMaj output ) {
output.reshape(3, H.numCols);
int stride = H.numCols;
for (int col = 0; col < H.numCols; col++) {
// This column in H
double h1 = H.data[col], h2 = H.data[col + stride], h3 = H.data[col + 2*stride];
output.data[col] = h1*stdX + h3*meanX;
output.data[col + stride] = h2*stdY + h3*meanY;
output.data[col + 2*stride] = h3;
}
}
public void apply( Point2D_F64 p, Point2D_F64 output ) {
output.x = (p.x - meanX)/stdX;
output.y = (p.y - meanY)/stdY;
}
public void apply( Point3D_F64 p, Point3D_F64 output ) {
output.x = (p.x - p.z*meanX)/stdX;
output.y = (p.y - p.z*meanY)/stdY;
}
/**
* C* = H'*C*H
*/
public void apply( ConicGeneral_F64 p, ConicGeneral_F64 output ) {
DMatrixRMaj C = UtilCurves_F64.convert(p, (DMatrixRMaj)null);
DMatrixRMaj Hinv = matrixInv(null);
DMatrixRMaj CP = new DMatrixRMaj(3, 3);
PerspectiveOps.multTranA(Hinv, C, Hinv, CP);
UtilCurves_F64.convert(CP, output);
}
/**
* Apply transform to conic in 3x3 matrix format.
*/
public void apply( DMatrix3x3 C, DMatrix3x3 output ) {
DMatrix3x3 Hinv = matrixInv3(work);
PerspectiveOps.multTranA(Hinv, C, Hinv, output);
}
public void remove( Point2D_F64 p, Point2D_F64 output ) {
output.x = p.x*stdX + meanX;
output.y = p.y*stdY + meanY;
}
public void remove( Point3D_F64 p, Point3D_F64 output ) {
output.x = p.x*stdX + p.z*meanX;
output.y = p.y*stdY + p.z*meanY;
}
public void remove( ConicGeneral_F64 p, ConicGeneral_F64 output ) {
DMatrixRMaj C = UtilCurves_F64.convert(p, (DMatrixRMaj)null);
DMatrixRMaj H = matrix(null);
DMatrixRMaj CP = new DMatrixRMaj(3, 3);
PerspectiveOps.multTranA(H, C, H, CP);
UtilCurves_F64.convert(CP, output);
}
public void remove( DMatrix3x3 C, DMatrix3x3 output ) {
DMatrix3x3 H = matrix3(work);
PerspectiveOps.multTranA(H, C, H, output);
}
public DMatrixRMaj matrix( @Nullable DMatrixRMaj M ) {
if (M == null)
M = new DMatrixRMaj(3, 3);
else
M.reshape(3, 3);
M.set(0, 0, 1.0/stdX);
M.set(1, 1, 1.0/stdY);
M.set(0, 2, -meanX/stdX);
M.set(1, 2, -meanY/stdY);
M.set(2, 2, 1);
return M;
}
public DMatrixRMaj matrixInv( @Nullable DMatrixRMaj M ) {
if (M == null)
M = new DMatrixRMaj(3, 3);
else
M.reshape(3, 3);
M.set(0, 0, stdX);
M.set(1, 1, stdY);
M.set(0, 2, meanX);
M.set(1, 2, meanY);
M.set(2, 2, 1);
return M;
}
public DMatrix3x3 matrix3( @Nullable DMatrix3x3 M ) {
if (M == null)
M = new DMatrix3x3();
else {
M.a21 = 0;
M.a31 = 0;
M.a31 = 0;
}
M.a11 = 1.0/stdX;
M.a12 = 1.0/stdY;
M.a13 = -meanX/stdX;
M.a23 = -meanY/stdY;
M.a22 = 1;
return M;
}
public DMatrix3x3 matrixInv3( DMatrix3x3 M ) {
if (M == null)
M = new DMatrix3x3();
else {
M.a21 = 0;
M.a31 = 0;
M.a31 = 0;
}
M.a11 = stdX;
M.a12 = stdY;
M.a13 = meanX;
M.a23 = meanY;
M.a22 = 1;
return M;
}
public boolean isEquals( NormalizationPoint2D a, double tol ) {
if (Math.abs(a.meanX - meanX) > tol)
return false;
if (Math.abs(a.meanY - meanY) > tol)
return false;
if (Math.abs(a.stdX - stdX) > tol)
return false;
if (Math.abs(a.stdY - stdY) > tol)
return false;
return true;
}
}