Method pixelToCoord

pdal/private/gdal/Raster.cpp:450–476

Convert a pixel X/Y position to a position in a coord. reference system. \param col Pixel column. \param row Pixel row. \param[out] output Array of 2 doubles with of the position in a CRS. */

  \param[out] output  Array of 2 doubles with of the position in a CRS.
*/
void Raster::pixelToCoord(int col, int row, std::array<double, 2>& output) const
{
    /**
    double *xform = const_cast<double *>(m_forwardTransform.data());
    GDALApplyGeoTransform(xform, col, row, &output[0], &output[1]);
    **/

    // from http://gis.stackexchange.com/questions/53617/how-to-find-lat-lon-values-for-every-pixel-in-a-geotiff-file
    double c = m_forwardTransform[0];
    double a = m_forwardTransform[1];
    double b = m_forwardTransform[2];
    double f = m_forwardTransform[3];
    double d = m_forwardTransform[4];
    double e = m_forwardTransform[5];

    //ABELL - Not sure why this is right.  You can think of this like:
    //   output[0] = a * (col + .5) + b * (row + .5) + c;
    //   output[1] = d * (col + .5) + e * (row + .5) + f;
    //   Is there some reason why you want to "move" the points in the raster
    //   to a location between the rows/columns?  Seems that you would just
    //   use 'c' and 'f' to shift everything a half-row and half-column if
    //   that's what you wanted.
    //   Also, this isn't what GDALApplyGeoTransform does.  And why aren't
    //   we just calling GDALApplyGeoTransform?
    output[0] = a*col + b*row + a*0.5 + b*0.5 + c;
    output[1] = d*col + e*row + d*0.5 + e*0.5 + f;
}


/**