Exemplo n.º 1
0
void smf_mapbounds( int fast, Grp *igrp,  int size, const char *system,
                    AstFrameSet *spacerefwcs, int alignsys, int *lbnd_out,
                    int *ubnd_out, AstFrameSet **outframeset, int *moving,
                    smfBox ** boxes, fts2Port fts_port, int *status ) {

  /* Local Variables */
  AstSkyFrame *abskyframe = NULL; /* Output Absolute SkyFrame */
  int actval;           /* Number of parameter values supplied */
  AstMapping *bolo2map = NULL; /* Combined mapping bolo->map
                                  coordinates, WCS->GRID Mapping from
                                  input WCS FrameSet */
  smfBox *box = NULL;          /* smfBox for current file */
  smfData *data = NULL;        /* pointer to  SCUBA2 data struct */
  double dlbnd[ 2 ];    /* Floating point lower bounds for output map */
  drcntrl_bits drcntrl_mask = 0;/* Mask to use for DRCONTROL on this instrument */
  double dubnd[ 2 ];    /* Floating point upper bounds for output map */
  AstMapping *fast_map = NULL; /* Mapping from tracking to absolute map coords */
  smfFile *file = NULL;        /* SCUBA2 data file information */
  int first;                   /* Is this the first good subscan ? */
  AstFitsChan *fitschan = NULL;/* Fits channels to construct WCS header */
  AstFrameSet *fs = NULL;      /* A general purpose FrameSet pointer */
  smfHead *hdr = NULL;         /* Pointer to data header this time slice */
  int i;                       /* Loop counter */
  dim_t j;                     /* Loop counter */
  AstSkyFrame *junksky = NULL; /* Unused SkyFrame argument */
  dim_t k;                     /* Loop counter */
  int lbnd0[ 2 ];              /* Defaults for LBND parameter */
  double map_pa=0;             /* Map PA in output coord system (rads) */
  dim_t maxloop;               /* Number of times to go round the time slice loop */
  dim_t nbadt  = 0;            /* Number of bad time slices */
  dim_t ngoodt = 0;            /* Number of good time slices */
  double par[7];               /* Projection parameters */
  double shift[ 2 ];           /* Shifts from PIXEL to GRID coords */
  AstMapping *oskymap = NULL;  /* Mapping celestial->map coordinates,
                                  Sky <> PIXEL mapping in output
                                  FrameSet */
  AstSkyFrame *oskyframe = NULL;/* Output SkyFrame */
  char *refsys = NULL;         /* Sky system from supplied reference FrameSet */
  dim_t textreme[4];           /* Time index corresponding to minmax TCS posn */
  AstFrame *skyin = NULL;      /* Sky Frame in input FrameSet */
  double skyref[ 2 ];          /* Values for output SkyFrame SkyRef attribute */
  struct timeval tv1;          /* Timer */
  struct timeval tv2;          /* Timer */
  AstMapping *tmap;            /* Temporary Mapping */
  int trim;                    /* Trim borders of bad pixels from o/p image? */
  int ubnd0[ 2 ];              /* Defaults for UBND parameter */
  double x_array_corners[4];   /* X-Indices for corner bolos in array */
  double x_map[4];             /* Projected X-coordinates of corner bolos */
  double y_array_corners[4];   /* Y-Indices for corner pixels in array */
  double y_map[4];             /* Projected X-coordinates of corner bolos */

  /* Main routine */
  if (*status != SAI__OK) return;

  /* Start a timer to see how long this takes */
  smf_timerinit( &tv1, &tv2, status );

  /* Initialize pointer to output FrameSet and moving-source flag */
  *outframeset = NULL;
  *moving = 0;

  /* initialize double precision output bounds and the proj pars */
  for( i = 0; i < 7; i++ ) par[ i ] = AST__BAD;
  dlbnd[ 0 ] = VAL__MAXD;
  dlbnd[ 1 ] = VAL__MAXD;
  dubnd[ 0 ] = VAL__MIND;
  dubnd[ 1 ] = VAL__MIND;

  /* If we have a supplied reference WCS we can use that directly
     without having to calculate it from the data. Replace the requested
     system with the system from the reference FrameSet (take a copy of the
     string since astGetC may re-use its buffer). */
  if (spacerefwcs) {
     oskyframe = astGetFrame( spacerefwcs, AST__CURRENT );
     int nc = 0;
     refsys = astAppendString( NULL, &nc, astGetC( oskyframe, "System" ) );
     system = refsys;
  }

  /* Create array of returned smfBox structures and store a pointer
     to the next one to be initialised. */
  *boxes = astMalloc( sizeof( smfBox ) * size );
  box = *boxes;

  astBegin;

  /* Loop over all files in the Grp */
  first = 1;
  for( i=1; i<=size; i++, box++ ) {

    /* Initialise the spatial bounds of section of the the output cube that is
       contributed to by the current ionput file. */
    box->lbnd[ 0 ] = VAL__MAXD;
    box->lbnd[ 1 ] = VAL__MAXD;
    box->ubnd[ 0 ] = VAL__MIND;
    box->ubnd[ 1 ] = VAL__MIND;

    /* Read data from the ith input file in the group */
    smf_open_file( NULL, igrp, i, "READ", SMF__NOCREATE_DATA, &data, status );

    if (*status != SAI__OK) {
      msgSeti( "I", i );
      errRep( "smf_mapbounds", "Could not open data file no ^I.", status );
      break;
    } else {
      if( *status == SAI__OK ) {
        if( data->file == NULL ) {
          *status = SAI__ERROR;
          errRep( FUNC_NAME, "No smfFile associated with smfData.",
                  status );
          break;

        } else if( data->hdr == NULL ) {
          *status = SAI__ERROR;
          errRep( FUNC_NAME, "No smfHead associated with smfData.",
                  status );
          break;

        } else if( data->hdr->fitshdr == NULL ) {
          *status = SAI__ERROR;
          errRep( FUNC_NAME, "No FITS header associated with smfHead.",
                  status );
          break;

        }
      }
    }

    /* convenience pointers */
    file = data->file;
    hdr = data->hdr;

    /* report name of the input file */
    smf_smfFile_msg( file, "FILE", 1, "<unknown>" );
    msgSeti("I", i);
    msgSeti("N", size);
    msgOutif(MSG__VERB, " ",
             "SMF_MAPBOUNDS: Processing ^I/^N ^FILE",
             status);

/* Check that there are 3 pixel axes. */
    if( data->ndims != 3 ) {
      smf_smfFile_msg( file, "FILE", 1, "<unknown>" );
      msgSeti( "NDIMS", data->ndims );
      *status = SAI__ERROR;
      errRep( FUNC_NAME, "^FILE has ^NDIMS pixel axes, should be 3.",
              status );
      break;
    }

    /* Check that the data dimensions are 3 (for time ordered data) */
    if( *status == SAI__OK ) {

      /* If OK Decide which detectors (GRID coord) to use for
         checking bounds, depending on the instrument in use. */

      switch( hdr->instrument ) {

      case INST__SCUBA2:
        drcntrl_mask = DRCNTRL__POSITION;
        /* 4 corner bolometers of the subarray */
        x_array_corners[0] = 1;
        x_array_corners[1] = 1;
        x_array_corners[2] = (data->dims)[0];
        x_array_corners[3] = (data->dims)[0];

        y_array_corners[0] = 1;
        y_array_corners[1] = (data->dims)[1];
        y_array_corners[2] = 1;
        y_array_corners[3] = (data->dims)[1];
        break;

      case INST__AZTEC:
        /* Rough guess for extreme bolometers around the edge */
        x_array_corners[0] = 22;
        x_array_corners[1] = 65;
        x_array_corners[2] = 73;
        x_array_corners[3] = 98;

        y_array_corners[0] = 1; /* Always 1 for AzTEC */
        y_array_corners[1] = 1;
        y_array_corners[2] = 1;
        y_array_corners[3] = 1;
        break;

      case INST__ACSIS:
        smf_find_acsis_corners( data, x_array_corners, y_array_corners,
                                status);
        break;

      default:
        *status = SAI__ERROR;
        errRep(FUNC_NAME, "Don't know how to calculate mapbounds for data created with this instrument", status);
      }
    }

    if( *status == SAI__OK) {
      size_t goodidx = SMF__BADSZT;

      /* Need to build up a frameset based on good telescope position.
         We can not assume that we the first step will be a good TCS position
         so we look for one. If we can not find anything we skip to the
         next file. */
      maxloop = (data->dims)[2];
      for (j=0; j<maxloop; j++) {
        JCMTState state = (hdr->allState)[j];
        if (state.jos_drcontrol >= 0 && state.jos_drcontrol & drcntrl_mask ) {
          /* bad TCS - so try again */
        } else {
          /* Good tcs */
          goodidx = j;
          break;
        }
      }

      if (goodidx == SMF__BADSZT) {
        smf_smfFile_msg( data->file, "FILE", 1, "<unknown>");
        msgOutif( MSG__QUIET, "", "No good telescope positions found in file ^FILE. Ignoring",
                  status );
        smf_close_file( NULL, &data, status );
        continue;
      }

      /* If we are dealing with the first good file, create the output
         SkyFrame. */
      if( first ) {
        first = 0;

        /* Create output SkyFrame if it has not come from a reference */
        if ( oskyframe == NULL ) {

          /* smf_tslice_ast only needs to get called once to set up framesets */
          if( hdr->wcs == NULL ) {
            smf_tslice_ast( data, goodidx, 1, fts_port, status);
          }

          /* Retrieve input SkyFrame */
          skyin = astGetFrame( hdr->wcs, AST__CURRENT );

          smf_calc_skyframe( skyin, system, hdr, alignsys, &oskyframe, skyref,
                             moving, status );

          /* Get the orientation of the map vertical within the output celestial
             coordinate system. This is derived form the MAP_PA FITS header, which
             gives the orientation of the map vertical within the tracking system. */
          map_pa = smf_calc_mappa( hdr, system, skyin, status );

          /* Provide a sensible default for the pixel size based on wavelength */
          par[4] = smf_calc_telres( hdr->fitshdr, status );
          par[4] *= AST__DD2R/3600.0;
          par[5] = par[4];

          /* Calculate the projection parameters. We do not enable autogrid determination
             for SCUBA-2 so we do not need to obtain all the data before calculating
             projection parameters. */
          smf_get_projpar( oskyframe, skyref, *moving, 0, 0, NULL, 0,
                           map_pa, par, NULL, NULL, status );

          if (skyin) skyin = astAnnul( skyin );

        /* If the output skyframe has been supplied, we still need to
           determine whether the source is moving or not, and set the
           reference position. */
        } else {

          /* smf_tslice_ast only needs to get called once to set up framesets */
          if( hdr->wcs == NULL ) {
            smf_tslice_ast( data, goodidx, 1, fts_port, status);
          }

          /* Retrieve input SkyFrame */
          skyin = astGetFrame( hdr->wcs, AST__CURRENT );
          smf_calc_skyframe( skyin, system, hdr, alignsys, &junksky, skyref,
                             moving, status );

          /* Store the sky reference position. If the target is moving,
             ensure the returned SkyFrame represents offsets from the
             reference position rather than absolute coords. */
          astSetD( oskyframe, "SkyRef(1)", skyref[ 0 ] );
          astSetD( oskyframe, "SkyRef(2)", skyref[ 1 ] );
          if( *moving ) astSet( oskyframe, "SkyRefIs=Origin" );

          /* Ensure the Epoch attribute in the map is set to the date of
             the first data in the map, rather than the date in supplied
             reference WCS. */
          astSetD( oskyframe, "Epoch", astGetD( junksky, "Epoch" ) );
        }

        if ( *outframeset == NULL && oskyframe != NULL && (*status == SAI__OK)){
          /* Now created a spatial Mapping. Use the supplied reference frameset
             if supplied */
          if (spacerefwcs) {
            oskymap = astGetMapping( spacerefwcs, AST__BASE, AST__CURRENT );
          } else {
            /* Now populate a FitsChan with FITS-WCS headers describing
               the required tan plane projection. The longitude and
               latitude axis types are set to either (RA,Dec) or (AZ,EL)
               to get the correct handedness. */
            fitschan = astFitsChan ( NULL, NULL, " " );
            smf_makefitschan( astGetC( oskyframe, "System"), &(par[0]),
                              &(par[2]), &(par[4]), par[6], fitschan, status );
            astClear( fitschan, "Card" );
            fs = astRead( fitschan );

            /* Extract the output PIXEL->SKY Mapping. */
            oskymap = astGetMapping( fs, AST__BASE, AST__CURRENT );

            /* Tidy up */
            fs = astAnnul( fs );
          }

          /* Create the output FrameSet */
          *outframeset = astFrameSet( astFrame(2, "Domain=GRID"), " " );

          /* Now add the SkyFrame to it */
          astAddFrame( *outframeset, AST__BASE, oskymap, oskyframe );

          /* Now add a POLANAL Frame if required (i.e. if the input time
             series are POL-2 Q/U values). */
          smf_addpolanal( *outframeset, hdr, status );

          /* Invert the oskymap mapping */
          astInvert( oskymap );

        } /* End WCS FrameSet construction */
      }

      /* Get a copy of the output SkyFrame and ensure it represents
         absolute coords rather than offset coords. */
      abskyframe = astCopy( oskyframe );
      astClear( abskyframe, "SkyRefIs" );
      astClear( abskyframe, "AlignOffset" );

      maxloop = (data->dims)[2];
      if (fast) {
        /* For scan map we scan through looking for largest telescope moves.
           For dream/stare we just look at the start and end time slices to
           account for sky rotation. */

        if (hdr->obsmode != SMF__OBS_SCAN) {
          textreme[0] = 0;
          textreme[1] = (data->dims)[2] - 1;
          maxloop = 2;

        } else {
          const char *tracksys;
          double *ac1list, *ac2list, *bc1list, *bc2list, *p1, *p2, *p3, *p4;
          double flbnd[4], fubnd[4];
          JCMTState state;

          /* If the output and tracking systems are different, get a
             Mapping between them. */
          tracksys = sc2ast_convert_system( (hdr->allState)[goodidx].tcs_tr_sys,
                                            status );
          if( strcmp( system, tracksys ) ) {
             AstSkyFrame *tempsf = astCopy( abskyframe );
             astSetC( tempsf, "System", tracksys );
             AstFrameSet *tempfs = astConvert( tempsf, abskyframe, "" );
             tmap = astGetMapping( tempfs, AST__BASE, AST__CURRENT );
             fast_map = astSimplify( tmap );
             tmap = astAnnul( tmap );
             tempsf = astAnnul( tempsf );
             tempfs = astAnnul( tempfs );
          } else {
             fast_map = NULL;
          }

          /* Copy all ac1/2 positions into two array, and transform them
             from tracking to absolute output sky coords. */
          ac1list = astMalloc( maxloop*sizeof( *ac1list ) );
          ac2list = astMalloc( maxloop*sizeof( *ac2list ) );
          if( *status == SAI__OK ) {
             p1 = ac1list;
             p2 = ac2list;
             for( j = 0; j < maxloop; j++ ) {
                state = (hdr->allState)[ j ];
                *(p1++) = state.tcs_tr_ac1;
                *(p2++) = state.tcs_tr_ac2;
             }
             if( fast_map ) astTran2( fast_map, maxloop, ac1list, ac2list, 1,
                                      ac1list, ac2list );
          }

          /* If the target is moving, we need to adjust these ac1/2 values
             to represent offsets from the base position. */
          if( *moving ) {

          /* Copy all bc1/2 positions into two arrays. */
             bc1list = astMalloc( maxloop*sizeof( *bc1list ) );
             bc2list = astMalloc( maxloop*sizeof( *bc2list ) );
             if( *status == SAI__OK ) {
                p1 = bc1list;
                p2 = bc2list;

                for( j = 0; j < maxloop; j++ ) {
                   state = (hdr->allState)[ j ];
                   *(p1++) = state.tcs_tr_bc1;
                   *(p2++) = state.tcs_tr_bc2;
                }

                /* Transform them from tracking to absolute output sky coords. */
                if( fast_map ) astTran2( fast_map, maxloop, bc1list, bc2list,
                                         1, bc1list, bc2list );

                /* Replace each ac1/2 position with the offsets from the
                   corresponding base position. */
                p1 = bc1list;
                p2 = bc2list;
                p3 = ac1list;
                p4 = ac2list;
                for( j = 0; j < maxloop; j++ ) {
                  smf_offsets( *(p1++), *(p2++), p3++, p4++, status );
                }
             }

             /* We no longer need the base positions. */
             bc1list = astFree( bc1list );
             bc2list = astFree( bc2list );
          }

          /* Transform the ac1/2 position from output sky coords to
             output pixel coords. */
          astTran2( oskymap, maxloop, ac1list, ac2list, 1, ac1list, ac2list );

          /* Find the bounding box containing these pixel coords and the
             time slices at which the boresight touches each edge of this
             box. */
          flbnd[ 0 ] = VAL__MAXD;
          flbnd[ 1 ] = VAL__MAXD;
          fubnd[ 0 ] = VAL__MIND;
          fubnd[ 1 ] = VAL__MIND;
          for( j = 0; j < 4; j++ ) textreme[ j ] = (dim_t) VAL__BADI;

          if( *status == SAI__OK ) {
             p1 = ac1list;
             p2 = ac2list;
             for( j = 0; j < maxloop; j++,p1++,p2++ ) {
                if( *p1 != VAL__BADD && *p2 != VAL__BADD ){

                   if ( *p1 < flbnd[0] ) { flbnd[0] = *p1; textreme[0] = j; }
                   if ( *p2 < flbnd[1] ) { flbnd[1] = *p2; textreme[1] = j; }
                   if ( *p1 > fubnd[0] ) { fubnd[0] = *p1; textreme[2] = j; }
                   if ( *p2 > fubnd[1] ) { fubnd[1] = *p2; textreme[3] = j; }
                }
             }
          }

          maxloop = 4;
          msgSetd("X1", textreme[0]);
          msgSetd("X2", textreme[1]);
          msgSetd("X3", textreme[2]);
          msgSetd("X4", textreme[3]);
          msgOutif( MSG__DEBUG, " ",
                    "Extrema time slices are ^X1, ^X2, ^X3 and ^X4",
                    status);

          ac1list = astFree( ac1list );
          ac2list = astFree( ac2list );

        }
      }

      /* Get the astrometry for all the relevant time slices in this data file */
      for( j=0; j<maxloop; j++ ) {
        dim_t ts;  /* Actual time slice to use */

        /* if we are doing the fast loop, we need to read the time slice
           index from textreme. Else we just use the index */
        if (fast) {
          /* get the index but make sure it is good */
          ts = textreme[j];
          if (ts == (dim_t)VAL__BADI) continue;
        } else {
          ts = j;
        }
        /* Calculate the bolo to map-pixel transformation for this tslice */
        bolo2map = smf_rebin_totmap( data, ts, abskyframe, oskymap,
                                     *moving, fts_port, status );

        if ( *status == SAI__OK ) {
          /* skip if we did not get a mapping this time round */
          if (!bolo2map) continue;

          /* Check corner pixels in the array for their projected extent
             on the sky to set the pixel bounds */
          astTran2( bolo2map, 4, x_array_corners, y_array_corners, 1,
                    x_map, y_map );

          /* Update min/max for this time slice */
          for( k=0; k<4; k++ ) {

            if( x_map[k] != AST__BAD && y_map[k] != AST__BAD ) {
              if( x_map[k] < dlbnd[0] ) dlbnd[0] = x_map[k];
              if( y_map[k] < dlbnd[1] ) dlbnd[1] = y_map[k];
              if( x_map[k] > dubnd[0] ) dubnd[0] = x_map[k];
              if( y_map[k] > dubnd[1] ) dubnd[1] = y_map[k];

              if( x_map[k] < box->lbnd[0] ) box->lbnd[0] = x_map[k];
              if( y_map[k] < box->lbnd[1] ) box->lbnd[1] = y_map[k];
              if( x_map[k] > box->ubnd[0] ) box->ubnd[0] = x_map[k];
              if( y_map[k] > box->ubnd[1] ) box->ubnd[1] = y_map[k];

            } else if( *status == SAI__OK ) {
              *status = SAI__ERROR;
              errRep( FUNC_NAME, "Extreme positions are bad.", status );
              break;
            }
          }
        }
        /* Explicitly annul these mappings each time slice for reduced
           memory usage */
        if (bolo2map) bolo2map = astAnnul( bolo2map );
        if (fs) fs = astAnnul( fs );

        /* Break out of loop over time slices if bad status */
        if (*status != SAI__OK) goto CLEANUP;
      }

      /* Annul any remaining Ast objects before moving on to the next file */
      if (fs) fs = astAnnul( fs );
      if (bolo2map) bolo2map = astAnnul( bolo2map );
    }

    /* Close the data file */
    smf_close_file( NULL, &data, status);

    /* Break out of loop over data files if bad status */
    if (*status != SAI__OK) goto CLEANUP;
  }

  /* make sure we got values - should not be possible with good status */
  if (dlbnd[0] == VAL__MAXD || dlbnd[1] == VAL__MAXD) {
    if (*status == SAI__OK) {
      *status = SAI__ERROR;
      errRep( " ", "Unable to find any valid map bounds", status );
    }
  }

  if (nbadt > 0) {
    msgOutf( "", "   Processed %zu time slices to calculate bounds,"
             " of which %zu had bad telescope data and were skipped",
             status, (size_t)(ngoodt+nbadt), (size_t)nbadt );
  }

  /* If spatial reference wcs was supplied, store par values that result in
     no change to the pixel origin. */
  if( spacerefwcs ){
    par[ 0 ] = 0.5;
    par[ 1 ] = 0.5;
  }

  /* Need to re-align with the interim GRID coordinates */
  lbnd_out[0] = ceil( dlbnd[0] - par[0] + 0.5 );
  ubnd_out[0] = ceil( dubnd[0] - par[0] + 0.5 );
  lbnd_out[1] = ceil( dlbnd[1] - par[1] + 0.5 );
  ubnd_out[1] = ceil( dubnd[1] - par[1] + 0.5 );

  /* Do the same with the individual input file bounding boxes */
  box = *boxes;
  for (i = 1; i <= size; i++, box++) {
    box->lbnd[0] = ceil( box->lbnd[0] - par[0] + 0.5);
    box->ubnd[0] = ceil( box->ubnd[0] - par[0] + 0.5);
    box->lbnd[1] = ceil( box->lbnd[1] - par[1] + 0.5);
    box->ubnd[1] = ceil( box->ubnd[1] - par[1] + 0.5);
  }

  /* Apply a ShiftMap to the output FrameSet to re-align the GRID
     coordinates */
  shift[0] = 2.0 - par[0] - lbnd_out[0];
  shift[1] = 2.0 - par[1] - lbnd_out[1];
  astRemapFrame( *outframeset, AST__BASE, astShiftMap( 2, shift, " " ) );

  /* Set the dynamic defaults for lbnd/ubnd */
  lbnd0[ 0 ] = lbnd_out[ 0 ];
  lbnd0[ 1 ] = lbnd_out[ 1 ];
  parDef1i( "LBND", 2, lbnd0, status );

  ubnd0[ 0 ] = ubnd_out[ 0 ];
  ubnd0[ 1 ] = ubnd_out[ 1 ];
  parDef1i( "UBND", 2, ubnd0, status );

  parGet1i( "LBND", 2, lbnd_out, &actval, status );
  if( actval == 1 ) lbnd_out[ 1 ] = lbnd_out[ 0 ];

  parGet1i( "UBND", 2, ubnd_out, &actval, status );
  if( actval == 1 ) ubnd_out[ 1 ] = ubnd_out[ 0 ];

  /* Ensure the bounds are the right way round. */
  if( lbnd_out[ 0 ] > ubnd_out[ 0 ] ) {
    int itmp = lbnd_out[ 0 ];
    lbnd_out[ 0 ] = ubnd_out[ 0 ];
    ubnd_out[ 0 ] = itmp;
  }

  if( lbnd_out[ 1 ] > ubnd_out[ 1 ] ) {
    int itmp = lbnd_out[ 1 ];
    lbnd_out[ 1 ] = ubnd_out[ 1 ];
    ubnd_out[ 1 ] = itmp;
  }

  /* If borders of bad pixels are being trimmed from the output image,
     then do not allow the user-specified bounds to extend outside the
     default bounding box (since we know that the default bounding box
     encloses all available data). */
  parGet0l( "TRIM", &trim, status );
  if( trim ) {
     if( lbnd_out[ 0 ] < lbnd0[ 0 ] ) lbnd_out[ 0 ] = lbnd0[ 0 ];
     if( lbnd_out[ 1 ] < lbnd0[ 1 ] ) lbnd_out[ 1 ] = lbnd0[ 1 ];
     if( ubnd_out[ 0 ] > ubnd0[ 0 ] ) ubnd_out[ 0 ] = ubnd0[ 0 ];
     if( ubnd_out[ 1 ] > ubnd0[ 1 ] ) ubnd_out[ 1 ] = ubnd0[ 1 ];
  }

  /* Modify the returned FrameSet to take account of the new pixel origin. */
  shift[ 0 ] = lbnd0[ 0 ] - lbnd_out[ 0 ];
  shift[ 1 ] = lbnd0[ 1 ] - lbnd_out[ 1 ];
  if( shift[ 0 ] != 0.0 || shift[ 1 ] != 0.0 ) {
    astRemapFrame( *outframeset, AST__BASE, astShiftMap( 2, shift, " " ) );
  }

/* Report the pixel bounds of the cube. */
  if( *status == SAI__OK ) {
    msgOutif( MSG__NORM, " ", " ", status );
    msgSeti( "XL", lbnd_out[ 0 ] );
    msgSeti( "YL", lbnd_out[ 1 ] );
    msgSeti( "XU", ubnd_out[ 0 ] );
    msgSeti( "YU", ubnd_out[ 1 ] );
    msgOutif( MSG__NORM, " ", "   Output map pixel bounds: ( ^XL:^XU, ^YL:^YU )",
              status );

    if( ( ubnd_out[ 0 ] - lbnd_out[ 0 ] + 1 ) > MAX_DIM ||
        ( ubnd_out[ 1 ] - lbnd_out[ 1 ] + 1 ) > MAX_DIM ) {
      *status = SAI__ERROR;
      errRep( "", FUNC_NAME ": The map is too big. Check your list of input "
              "data files does not include widely separated observations.",
              status );
    }
  }

  /* If no error has occurred, export the returned FrameSet pointer from the
     current AST context so that it will not be annulled when the AST
     context is ended. Otherwise, ensure a null pointer is returned. */
  if( *status == SAI__OK ) {
    astExport( *outframeset );
  } else {
    *outframeset = astAnnul( *outframeset );
  }

  msgOutiff( SMF__TIMER_MSG, "",
             "Took %.3f s to calculate map bounds",
             status, smf_timerupdate( &tv1, &tv2, status ) );

  /* Clean Up */
 CLEANUP:
  if (*status != SAI__OK) {
    errRep(FUNC_NAME, "Unable to determine map bounds", status);
  }
  if (oskymap) oskymap  = astAnnul( oskymap );
  if (bolo2map) bolo2map = astAnnul( bolo2map );
  if (fitschan) fitschan = astAnnul( fitschan );

  if( data != NULL )
    smf_close_file( NULL, &data, status );

  refsys = astFree( refsys );

  astEnd;

}
Exemplo n.º 2
0
void findback( int *status ){
/*
*+
*  Name:
*     FINDBACK

*  Purpose:
*     Estimate the background in an NDF by removing small scale structure.

*  Language:
*     C

*  Type of Module:
*     ADAM A-task

*  Synopsis:
*     void findback( int *status );

*  Description:
*     This application uses spatial filtering to remove structure with a
*     scale size less than a specified size from a 1, 2, or 3 dimensional
*     NDF, thus producing an estimate of the local background within the NDF.
*
*     The algorithm proceeds as follows. A filtered form of the input data
*     is first produced by replacing every input pixel by the minimum of
*     the input values within a rectangular box centred on the pixel.
*     This filtered data is then filtered again, using a filter that
*     replaces every pixel value by the maximum value in a box centred on
*     the pixel. This produces an estimate of the lower envelope of the data,
*     but usually contains unacceptable sharp edges. In addition, this
*     filtered data has a tendency to hug the lower envelope of the
*     noise, thus under-estimating the true background of the noise-free
*     data. The first problem is minimised by smoothing the background
*     estimate using a filter that replaces every pixel value by the mean
*     of the values in a box centred on the pixel. The second problem
*     is minimised by estimating the difference between the input data
*     and the background estimate within regions well removed from any
*     bright areas. This difference is then extrapolated into the bright
*     source regions and used as a correction to the background estimate.
*     Specifically, the residuals between the input data and the initial
*     background estimate are first formed, and residuals which are more
*     than three times the RMS noise are set bad. The remaining residuals
*     are smoothed with a mean filter. This smoothing will replace a lot
*     of the bad values rejected above, but may not remove them all. Any
*     remaining bad values are estimated by linear interpolation between
*     the nearest good values along the first axis. The interpolated
*     residuals are then smoothed again using a mean filter, to get a
*     surface representing the bias in the initial background estimate.
*     This surface is finally added onto the initial background estimate
*     to obtain the output NDF.

*  Usage:
*     findback in out box

*  ADAM Parameters:
*     BOX() = _INTEGER (Read)
*        The dimensions of each of the filters, in pixels. Each value
*        should be odd (if an even value is supplied, the next higher odd
*        value will be used). The number of values supplied should not
*        exceed the number of significant (i.e. more than one element)
*        pixel axes in the input array. If any trailing values of 1 are
*        supplied, then each pixel value on the corresponding axes
*        will be fitted independently of its neighbours. For instance,
*        if the data array is 3-dimensional, and the third BOX value is 1,
*        then each x-y plane will be fitted independently of the neighbouring
*        planes. If the NDF has more than 1 pixel axis but only 1 value is
*        supplied, then the same value will be used for the both the first
*        and second pixel axes (a value of 1 will be assumed for the third
*        axis if the input array is 3-dimensional).
*     MSG_FILTER = _CHAR (Read)
*        Controls the amount of diagnostic information reported. This is the
*        standard messaging level. The default messaging level is NORM (2).
*        A value of NONE or 0 will suppress all screen output. VERB (3) will
*        indicate progress through the various stages of the algorithm. [NORM]
*     IN = NDF (Read)
*        The input NDF.
*     RMS = _DOUBLE (Read)
*        Specifies a value to use as the global RMS noise level in the
*        supplied data array. The suggested default value is the square root
*        of the mean of the values in the input NDF's Variance component.
*        If the NDF has no Variance component, the suggested default
*        is based on the differences between neighbouring pixel values,
*        measured over the entire input NDF. If multiple slices within the
*        NDF are to be processed independently (see parameter BOX), it
*        may be more appropriate for a separate default RMS to be calculated
*        for each slice. This will normally be the case if the noise could
*        be different in each of the slices. In such cases a null (!) can
*        be supplied for the RMS parameter, which forces a separate
*        default RMS value to be found and used for each slice. Any
*        pixel-to-pixel correlation in the noise can result in these
*        defaults being too low.
*     SUB = _LOGICAL (Read)
*        If a TRUE value is supplied, the output NDF will contain the
*        difference between the supplied input data and the estimated
*        background. If a FALSE value is supplied, the output NDF will
*        contain the estimated background itself. [FALSE]
*     OUT = NDF (Write)
*        The output NDF containing either the estimated background, or the
*        background-subtracted input data, as specified by parameter SUB.

*  Notes:
*     - Smoothing cubes in 3 dimensions can be very slow.

*  Copyright:
*     Copyright (C) 2009 Science and Technology Facilities Council.
*     Copyright (C) 2006, 2007 Particle Physics & Astronomy Research Council.
*     All Rights Reserved.

*  Licence:
*     This program is free software; you can redistribute it and/or
*     modify it under the terms of the GNU General Public License as
*     published by the Free Software Foundation; either version 2 of
*     the License, or (at your option) any later version.
*
*     This program is distributed in the hope that it will be
*     useful, but WITHOUT ANY WARRANTY; without even the implied
*     warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
*     PURPOSE. See the GNU General Public License for more details.
*
*     You should have received a copy of the GNU General Public License
*     along with this program; if not, write to the Free Software
*     Foundation, Inc., 51 Franklin Street,Fifth Floor, Boston, MA
*     02110-1301, USA

*  Authors:
*     DSB: David S. Berry
*     TIMJ: Tim Jenness (JAC, Hawaii)
*     {enter_new_authors_here}

*  History:
*     13-SEP-2006 (DSB):
*        Original version.
*     19-MAR-2007 (DSB):
*        - Added parameters SUB and RMS.
*        - Fix bug that left the output NDF uninitialised if ILEVEL is set
*        non-zero.
*        - Use generic data type handling as in FINDCLUMPS.
*     14-JAN-2009 (TIMJ):
*        Use MERS for message filtering.
*     29-JUL-2009 (TIMJ):
*        Rename ILEVEL to MSG_FILTER
*     17-MAY-2011 (DSB):
*        Use sqrt rather than sqrtf when calculating RMS.
*     12-SEP-2011 (DSB):
*        Process slices in separate threads.
*     {enter_further_changes_here}

*-
*/

/* Local Variables: */
   CupidFindback0Data *job_data; /* Pointer to data for all jobs */
   CupidFindback0Data *pdata; /* Pointer to data for current job */
   Grp *grp;                 /* GRP identifier for configuration settings */
   ThrWorkForce *wf = NULL;  /* Pool of persistent worker threads */
   char dtype[ 21 ];         /* HDS data type for output NDF */
   char itype[ 21 ];         /* HDS data type to use when processing */
   double *ipv;              /* Pointer to Variance array */
   double *pd1;              /* Pointer to double precision input data */
   double *pd2;              /* Pointer to double precision output data */
   double rms;               /* Global rms error in data */
   double sum;               /* Sum of variances */
   float *pf1;               /* Pointer to single precision input data */
   float *pf2;               /* Pointer to single precision output data */
   int *old_status;          /* Pointer to original status value */
   int box[ 3 ];             /* Dimensions of each cell in pixels */
   int dim[ NDF__MXDIM ];    /* Dimensions of each NDF pixel axis */
   int el;                   /* Number of elements mapped */
   int i;                    /* Loop count */
   int indf1;                /* Identifier for input NDF */
   int indf2;                /* Identifier for output NDF */
   int islice;               /* Slice index */
   int iystep;               /* Index of slice in ydirection */
   int izstep;               /* Index of slice in z direction */
   int lbnd[ NDF__MXDIM ];   /* Lower pixel bounds of slice */
   int n;                    /* Number of values summed in "sum" */
   int ndim;                 /* Total number of pixel axes in NDF */
   int newalg;               /* Use experimental algorithm variations? */
   int nsdim;                /* Number of significant pixel axes in NDF */
   int nslice;               /* Number of slices to process */
   int nval;                 /* Number of values supplied */
   int nystep;               /* Number of independent y slices */
   int nzstep;               /* Number of slices in z direction */
   int sdim[ 3 ];            /* Dimensions of each significant NDF axis */
   int slice_dim[ 3 ];       /* Dimensions of each significant slice axis */
   int slice_lbnd[ 3 ];      /* Lower bounds of each significant slice axis */
   int slice_size;           /* Number of pixels in each slice */
   int state;                /* Parameter state */
   int sub;                  /* Output the background-subtracted input data? */
   int type;                 /* Integer identifier for data type */
   int ubnd[ NDF__MXDIM ];   /* Upper pixel bounds of slice */
   int var;                  /* Does i/p NDF have a Variance component? */
   size_t size;              /* Size of GRP group */
   void *ipd1;               /* Pointer to input Data array */
   void *ipd2;               /* Pointer to output Data array */
   void *ipdin;              /* Pointer to input Data array */
   void *ipdout;             /* Pointer to output Data array */

/* Abort if an error has already occurred. */
   if( *status != SAI__OK ) return;

/* Start an NDF context */
   ndfBegin();

/* Record the existing AST status pointer, and ensure AST uses the supplied
   status pointer instead. */
   old_status = astWatch( status );

/* Get an identifier for the input NDF. We use NDG (via kpg1_Rgndf)
   instead of calling ndfAssoc directly since NDF/HDS has problems with
   file names containing spaces, which NDG does not have. */
   kpg1Rgndf( "IN", 1, 1, "", &grp, &size, status );
   ndgNdfas( grp, 1, "READ", &indf1, status );
   grpDelet( &grp, status );

/* Get the pixel index bounds of the input NDF. */
   ndfBound( indf1, NDF__MXDIM, lbnd, ubnd, &ndim, status );

/* Identify and count the number of significant axes (i.e. axes spanning
   more than 1 pixel). Also record their dimensions. */
   nsdim = 0;
   for( i = 0; i < ndim; i++ ) {
      dim[ i ] = ubnd[ i ] - lbnd[ i ] + 1;
      if( dim[ i ] > 1 ) sdim[ nsdim++ ] = dim[ i ];
   }

/* If there are too many significant axes, report an error. */
   if( nsdim > 3 && *status == SAI__OK ) {
       *status = SAI__ERROR;
       ndfMsg( "N", indf1 );
       msgSeti( "NS", nsdim );
       errRep( "", "The NDF '^N' has ^NS significant pixel axes, but this"
               "application requires 1, 2 or 3.", status );
   }

/* Ensure we have 3 values in sdim (pad with trailings 1's if required). */
   if( nsdim < 3 ) sdim[ 2 ] = 1;
   if( nsdim < 2 ) sdim[ 1 ] = 1;

/* See if the output is to contain the background-subtracted data, or the
   background estimate itself. */
   parGet0l( "SUB", &sub, status );

/* Create the output by propagating everything except the Data and
   (if we are outputting the background itself) Variance arrays. */
   if( sub ) {
      ndfProp( indf1, "UNITS,AXIS,WCS,QUALITY,VARIANCE", "OUT", &indf2,
               status );
   } else {
      ndfProp( indf1, "UNITS,AXIS,WCS,QUALITY", "OUT", &indf2, status );
   }

   msgBlankif( MSG__VERB, status );

/* Get the dimensions of each of the filters, in pixels. If only one
   value is supplied, duplicate it as the second value if the second axis
   is significant. If fewer than 3 values were supplied, use 1 for the 3rd
   value (whether or not it is significant). This results in each plane
   being fitted independently of the adjacent planes by default. */
   parGet1i( "BOX", nsdim, box, &nval, status );
   if( *status != SAI__OK ) goto L999;
   if( nval < 2 ) box[ 1 ] = ( nsdim > 1 ) ? box[ 0 ] : 1;
   if( nval < 3 ) box[ 2 ] = 1;

/* Ensure box sizes are odd. */
   box[ 0 ] = 2*( box[ 0 ] / 2 ) + 1;
   box[ 1 ] = 2*( box[ 1 ] / 2 ) + 1;
   box[ 2 ] = 2*( box[ 2 ] / 2 ) + 1;

   msgOutiff( MSG__VERB, "", "Using box sizes [%d,%d,%d].", status,
              box[0], box[1], box[2]);

/* If any trailing axes have a cell size of 1, then we apply the algorithm
   independently to every pixel index on the trailing axes. First of all
   set things up assuming that there are no trailing axes with cell size
   of 1. */
   nystep = 1;
   nzstep = 1;
   slice_dim[ 0 ] = sdim[ 0 ];
   slice_dim[ 1 ] = sdim[ 1 ];
   slice_dim[ 2 ] = sdim[ 2 ];
   slice_lbnd[ 0 ] = lbnd[ 0 ];
   slice_lbnd[ 1 ] = lbnd[ 1 ];
   slice_lbnd[ 2 ] = lbnd[ 2 ];

/* If the 3rd pixel axis has a cell size of 1, arrange that each slice
   contains a single plane. */
   if( box[ 2 ] == 1 ) {
      nzstep = sdim[ 2 ];
      slice_dim[ 2 ] = 1;

/* If the 2nd pixel axis also has a cell size of 1, arrange that each slice
   contains a single row. */
      if( box[ 1 ] == 1 ) {
         nystep = sdim[ 1 ];
         slice_dim[ 1 ] = 1;
      }
   }

/* Determine the number of pixels in each independent slice. */
   slice_size = slice_dim[ 0 ]*slice_dim[ 1 ]*slice_dim[ 2 ];

/* Decide what numeric data type to use, and set the output NDF data type. */
   ndfMtype( "_REAL,_DOUBLE", indf1, indf1, "Data,Variance", itype,
             20, dtype, 20, status );
   if( !strcmp( itype, "_DOUBLE" ) ) {
      type = CUPID__DOUBLE;
   } else {
      type = CUPID__FLOAT;
   }

   ndfStype( dtype, indf2, "Data,Variance", status );

/* Map the input and output arrays. */
   ndfMap( indf1, "Data", itype, "READ", &ipdin, &el, status );
   ndfMap( indf2, "Data", itype, "WRITE", &ipdout, &el, status );

/* If the rms value is supplied on the command, there is no need to
   calculate a default value. */
   parState( "RMS", &state, status );
   if( state == PAR__GROUND ) {

/* Calculate the default RMS value. If the NDF has a Variance component
   it is the square root of the mean Variance value. Otherwise, it is found
   by looking at differences between adjacent pixel values in the Data
   component. */
      ndfState( indf1, "VARIANCE", &var, status );
      if( *status == SAI__OK && var ) {
         ndfMap( indf1, "VARIANCE", "_DOUBLE", "READ", (void *) &ipv, &el, status );

         sum = 0.0;
         n = 0;
         for( i = 0; i < el; i++ ) {
            if( ipv[ i ] != VAL__BADD ) {
               sum += ipv[ i ];
               n++;
            }
         }

         if( n > 0 ) {
            rms = sqrt( sum/n );

         } else {
            *status = SAI__ERROR;
            errRep( "", "The supplied data contains insufficient "
                    "good Variance values to continue.", status );
         }

      } else {
         ipv = NULL;
         rms = cupidRms( type, ipdin, el, sdim[ 0 ], status );
      }

/* Set the default RMS noise level. */
      parDef0d( "RMS", rms, status );
   }

/* Abort if an error has occurred. */
   if( *status != SAI__OK ) goto L999;

/* Get the RMS noise level. */
   parGet0d( "RMS", &rms, status );

/* Annul the error and use an RMS value of VAL__BAD if a null parameter
   value was supplied. This causes an independent default noise estimate to
   be used for each slice of the base NDF. */
   if( *status == PAR__NULL ) {
      errAnnul( status );
      rms = VAL__BADD;
   }

/* See if any experimental algorithm variations are to be used. */
   parGet0l( "NEWALG", &newalg, status );

/* Create a pool of worker threads. */
   wf = thrCreateWorkforce( thrGetNThread( "CUPID_THREADS", status ), status );

/* Get memory to hold a description of each job passed to a worker. There
   is one job for each slice. */
   nslice = nystep*nzstep;
   job_data = astMalloc( nslice*sizeof( *job_data ) );
   if( *status == SAI__OK ) {

/* Loop round all slices to be processed. */
      ipd1 = ipdin;
      ipd2 = ipdout;
      islice = 0;
      pdata = job_data;

      for( izstep = 0; izstep < nzstep ; izstep++ ) {

         slice_lbnd[ 1 ] = lbnd[ 1 ];

         for( iystep = 0; iystep < nystep; iystep++, islice++,pdata++ ) {

/* Store the information needed by the function (cupidFindback0) that
   does the work in a thread. */
            pdata->islice = islice;
            pdata->nslice = nslice;
            pdata->type = type;
            pdata->ndim = ndim;
            pdata->box[ 0 ] = box[ 0 ];
            pdata->box[ 1 ] = box[ 1 ];
            pdata->box[ 2 ] = box[ 2 ];
            pdata->rms = rms;
            pdata->ipd1 = ipd1;
            pdata->ipd2 = ipd2;
            pdata->slice_dim[ 0 ] = slice_dim[ 0 ];
            pdata->slice_lbnd[ 0 ] = slice_lbnd[ 0 ];
            pdata->slice_dim[ 1 ] = slice_dim[ 1 ];
            pdata->slice_lbnd[ 1 ] = slice_lbnd[ 1 ];
            pdata->slice_dim[ 2 ] = slice_dim[ 2 ];
            pdata->slice_lbnd[ 2 ] = slice_lbnd[ 2 ];
            pdata->newalg = newalg;
            pdata->slice_size = slice_size;

/* Submit a job to the workforce to process the current slice. */
            thrAddJob( wf, 0, pdata, cupidFindback0, 0, NULL, status );

/* Update pointers to the start of the next slice in the input and output
   arrays. */
            if( type == CUPID__FLOAT ) {
               ipd1 = ( (float *) ipd1 ) + slice_size;
               ipd2 = ( (float *) ipd2 ) + slice_size;
            } else {
               ipd1 = ( (double *) ipd1 ) + slice_size;
               ipd2 = ( (double *) ipd2 ) + slice_size;
            }

/* Increment the lower bound on the 2nd pixel axis. */
            slice_lbnd[ 1 ]++;
         }

/* Increment the lower bound on the 3rd pixel axis. */
         slice_lbnd[ 2 ]++;
      }

/* Wait until all jobs have finished. */
      thrWait( wf, status );
   }

/* The output currently holds the background estimate. If the user has
   requested that the output should hold the background-subtracted input
   data, then do the arithmetic now. */
   if( sub && *status == SAI__OK ) {
      if( type == CUPID__FLOAT ) {
         pf1 = (float *) ipdin;
         pf2 = (float *) ipdout;
         for( i = 0; i < el; i++, pf1++, pf2++ ) {
            if( *pf1 != VAL__BADR && *pf2 != VAL__BADR ) {
               *pf2 = *pf1 - *pf2;
            } else {
               *pf2 = VAL__BADR;
            }
         }

      } else {
         pd1 = (double *) ipdin;
         pd2 = (double *) ipdout;
         for( i = 0; i < el; i++, pd1++, pd2++ ) {
            if( *pd1 != VAL__BADD && *pd2 != VAL__BADD ) {
               *pd2 = *pd1 - *pd2;
            } else {
               *pd2 = VAL__BADD;
            }
         }

      }
   }

/* Tidy up */
L999:;
   msgBlankif( MSG__VERB, status );

/* Free workspace. */
   job_data = astFree( job_data );
   wf = thrDestroyWorkforce( wf );

/* Reinstate the original AST inherited status value. */
   astWatch( old_status );

/* End the NDF context */
   ndfEnd( status );

/* If an error has occurred, issue another error report identifying the
   program which has failed (i.e. this one). */
   if( *status != SAI__OK ) {
      errRep( "FINDBACK_ERR", "FINDBACK: Failed to find the background "
              "of an NDF.", status );
   }
}