ReseauDistortionMap.cpp

 
/* SPDX-License-Identifier: CC0-1.0 */
#include "ReseauDistortionMap.h"
#include "LeastSquares.h"
#include "BasisFunction.h"
#include "CameraFocalPlaneMap.h"
#include "Pvl.h"
#include "Statistics.h"
#include <float.h>
 
using namespace std;
namespace Isis {
  ReseauDistortionMap::ReseauDistortionMap(Camera *parent, Pvl &labels, const QString &fname) :
    CameraDistortionMap(parent, 1.0) {
    // Set up distortion coefficients
    Pvl mast(fname);
    PvlGroup dim = mast.findGroup("Dimensions");
    p_distortedLines = dim.findKeyword("DistortedLines");
    p_distortedSamps = dim.findKeyword("DistortedSamples");
    p_undistortedLines = dim.findKeyword("UndistortedLines");
    p_undistortedSamps = dim.findKeyword("UndistortedSamples");
    PvlGroup mastRes = mast.findGroup("MasterReseaus");
    PvlKeyword mline = mastRes.findKeyword("Line");
    PvlKeyword msamp = mastRes.findKeyword("Sample");
    p_numRes = mline.size();
    if(mline.size() != msamp.size()) {
      string msg = "The number of lines and samples for the master reseaus are";
      msg += "not equal, the data file may be bad";
      throw IException(IException::User, msg, _FILEINFO_);
    }
    for(int i = 0; i < p_numRes; i++) {
      p_mlines.push_back(toDouble(mline[i]));
      p_msamps.push_back(toDouble(msamp[i]));
    }
    p_pixelPitch = parent->PixelPitch();
 
    PvlGroup refRes = labels.findGroup("Reseaus", Pvl::Traverse);
    PvlKeyword rline = refRes.findKeyword("Line");
    PvlKeyword rsamp = refRes.findKeyword("Sample");
    if(rline.size() != rsamp.size()) {
      string msg = "The number of lines and samples for the refined reseaus are";
      msg += "not equal, the data file may be bad";
      throw IException(IException::User, msg, _FILEINFO_);
    }
    for(int i = 0; i < p_numRes; i++) {
      p_rlines.push_back(toDouble(rline[i]));
      p_rsamps.push_back(toDouble(rsamp[i]));
    }
    if(p_mlines.size() != p_rlines.size()) {
      string msg = "The number of master reseaus and refined reseaus";
      msg += "do not appear to be equal";
      throw IException(IException::User, msg, _FILEINFO_);
    }
  }

  bool ReseauDistortionMap::SetFocalPlane(const double dx,
                                          const double dy) {
    p_focalPlaneX = dx;
    p_focalPlaneY = dy;
 
    // Convert distorted x,y position to a sample, line position
    double focalSamp = dx / p_pixelPitch +
                       p_camera->FocalPlaneMap()->DetectorSampleOrigin();
    double focalLine = dy / p_pixelPitch +
                       p_camera->FocalPlaneMap()->DetectorLineOrigin();
 
    // Find distance from input point to all nominal reseaus
    double distances[p_numRes], wt[5];
    int closepts[5];
    double ldiffsq, sdiffsq;
    for(int i = 0; i < p_numRes; i++) {
      sdiffsq = (focalSamp - p_rsamps[i]) * (focalSamp - p_rsamps[i]);
      ldiffsq = (focalLine - p_rlines[i]) * (focalLine - p_rlines[i]);
      distances[i]  =  ldiffsq + sdiffsq;
    }
 
    // Get 5 closest reseaus to the input point
    for(int ifpt = 0;  ifpt < 5;  ifpt++)  {
      int imin = 0;
      for(int i = 0;  i < p_numRes;  i++)  {
        if(distances[i] < distances[imin])  imin = i;
      }
      closepts[ifpt] = imin;
      wt[ifpt] = distances[imin];
      distances[imin] = DBL_MAX;  /* Arbitrary big number */
    }
 
    // Make sure 5 closest points are not colinear
    Statistics lstats, sstats;
    double line[5], samp[5];
    for(int k = 0; k < 5; k++) {
      line[k] = p_rlines[closepts[k]];
      samp[k] = p_rsamps[closepts[k]];
    }
    lstats.AddData(line, 5);
    sstats.AddData(samp, 5);
 
    // If they are colinear, return false
    if(lstats.StandardDeviation() < 10.0 || sstats.StandardDeviation() < 10.0) {
      return false;
    }
 
    // Weight each of the 5 closest points, and solve the system of equations
    if(wt[0] > 0.0)   {
      double scale = wt[0];
      double rfitlines[5], rfitsamps[5], mfitlines[5], mfitsamps[5];
      for(int ifpt = 0;  ifpt < 5; ifpt++)   {
        int index = closepts[ifpt];
        rfitlines[ifpt] = p_rlines[index];
        rfitsamps[ifpt] = p_rsamps[index];
        mfitlines[ifpt] = p_mlines[index];
        mfitsamps[ifpt] = p_msamps[index];
        wt[ifpt] = scale / wt[ifpt];
      }
      BasisFunction bsX("bilinearInterpX", 3, 3);
      BasisFunction bsY("bilinearInterpY", 3, 3);
      LeastSquares lsqX(bsX);
      LeastSquares lsqY(bsY);
 
      vector<double> known;
      known.resize(3);
      for(int i = 0; i < 5; i++) {
        known[0] = 1.0;
        known[1] = rfitsamps[i];
        known[2] = rfitlines[i];
        lsqX.AddKnown(known, mfitsamps[i], wt[i]);
        lsqY.AddKnown(known, mfitlines[i], wt[i]);
      }
      lsqX.Solve();
      lsqY.Solve();
 
      known[1] = focalSamp;
      known[2] = focalLine;
 
      // Test to make sure the point is inside of the image
      double undistortedFocalSamp = lsqX.Evaluate(known);
      double undistortedFocalLine = lsqY.Evaluate(known);
      if(undistortedFocalSamp < 0.5) return false;
      if(undistortedFocalLine < 0.5) return false;
      if(undistortedFocalSamp > p_undistortedSamps + 0.5) return false;
      if(undistortedFocalLine > p_undistortedLines + 0.5) return false;
 
      // Convert undistorted sample, line position to an x,y position
      p_undistortedFocalPlaneX = (undistortedFocalSamp - p_undistortedSamps
                                  / 2.0) * p_pixelPitch;
      p_undistortedFocalPlaneY = (undistortedFocalLine - p_undistortedLines
                                  / 2.0) * p_pixelPitch;
    }
    else   { // If the point passed in is a reseau...
      int index = closepts[0];
      p_undistortedFocalPlaneX = (p_msamps[index] - p_undistortedSamps / 2.0)
                                 * p_pixelPitch;
      p_undistortedFocalPlaneY = (p_mlines[index] - p_undistortedLines / 2.0)
                                 * p_pixelPitch;
    }
    return true;
  }

  bool ReseauDistortionMap::SetUndistortedFocalPlane(const double ux,
      const double uy) {
    p_undistortedFocalPlaneX = ux;
    p_undistortedFocalPlaneY = uy;
 
    // Convert undistorted values to sample, line positions
    double undistortedFocalSamp = ux / p_pixelPitch + p_undistortedSamps / 2.0;
    double undistortedFocalLine = uy / p_pixelPitch + p_undistortedLines / 2.0;
 
    // Find distance from input point to all nominal reseaus
    double distances[p_numRes], wt[5];
    int closepts[5];
    double ldiffsq, sdiffsq;
    for(int i = 0; i < p_numRes; i++) {
      sdiffsq = (undistortedFocalSamp - p_msamps[i]) *
                (undistortedFocalSamp - p_msamps[i]);
      ldiffsq = (undistortedFocalLine - p_mlines[i]) *
                (undistortedFocalLine - p_mlines[i]);
      distances[i]  =  ldiffsq + sdiffsq;
    }
 
    // Get 5 closest reseaus to the input point
    for(int ifpt = 0;  ifpt < 5;  ifpt++)  {
      int imin = 0;
      for(int i = 0;  i < p_numRes;  i++)  {
        if(distances[i] < distances[imin])  imin = i;
      }
      closepts[ifpt] = imin;
      wt[ifpt] = distances[imin];
      distances[imin] = DBL_MAX;  /* Arbitrary big number */
    }
 
    // Make sure 5 closest points are not colinear
    Statistics lstats, sstats;
    double line[5], samp[5];
    for(int k = 0; k < 5; k++) {
      line[k] = p_rlines[closepts[k]];
      samp[k] = p_rsamps[closepts[k]];
    }
    lstats.AddData(line, 5);
    sstats.AddData(samp, 5);
    // If they are colinear, return false
    if(lstats.StandardDeviation() < 10.0 || sstats.StandardDeviation() < 10.0) {
      return false;
    }
 
    // Weight each of the 5 closest points and solve the system of equations
    if(wt[0] > 0.0)   {
      double scale = wt[0];
      double rfitlines[5], rfitsamps[5], mfitlines[5], mfitsamps[5];
      for(int ifpt = 0;  ifpt < 5; ifpt++)   {
        int index = closepts[ifpt];
        mfitlines[ifpt] = p_mlines[index];
        mfitsamps[ifpt] = p_msamps[index];
        rfitlines[ifpt] = p_rlines[index];
        rfitsamps[ifpt] = p_rsamps[index];
        wt[ifpt] = scale / wt[ifpt];
      }
      BasisFunction bsX("bilinearInterpX", 3, 3);
      BasisFunction bsY("bilinearInterpY", 3, 3);
      LeastSquares lsqX(bsX);
      LeastSquares lsqY(bsY);
 
      vector<double> known;
      known.resize(3);
      for(int i = 0; i < 5; i++) {
        known[0] = 1.0;
        known[1] = mfitsamps[i];
        known[2] = mfitlines[i];
        lsqX.AddKnown(known, rfitsamps[i], wt[i]);
        lsqY.AddKnown(known, rfitlines[i], wt[i]);
      }
      lsqX.Solve();
      lsqY.Solve();
 
      known[1] = undistortedFocalSamp;
      known[2] = undistortedFocalLine;
 
      // Test points to make sure they are in the image
      double distortedFocalSamp = lsqX.Evaluate(known);
      double distortedFocalLine = lsqY.Evaluate(known);
      if(distortedFocalSamp < 0.5) return false;
      if(distortedFocalLine < 0.5) return false;
      if(distortedFocalSamp > p_undistortedSamps + 0.5) return false;
      if(distortedFocalLine > p_undistortedLines + 0.5) return false;
 
      // Convert distorted sample, line position back to an x,y position
      p_focalPlaneX = (distortedFocalSamp -
                       p_camera->FocalPlaneMap()->DetectorSampleOrigin()) * p_pixelPitch;
      p_focalPlaneY = (distortedFocalLine -
                       p_camera->FocalPlaneMap()->DetectorLineOrigin()) * p_pixelPitch;
    }
 
    else   { // If the point passed in is a reseau...
      int index = closepts[0];
      p_focalPlaneX = (p_rsamps[index] -
                       p_camera->FocalPlaneMap()->DetectorSampleOrigin()) * p_pixelPitch;
      p_focalPlaneY = (p_rlines[index] -
                       p_camera->FocalPlaneMap()->DetectorLineOrigin()) * p_pixelPitch;
    }
    return true;
  }
 
} // end namespace isis