Return to Post-BCD pages
 
Manual Contents
 
Getting Started
 
Input Files
 
Background Matching (overlap.pl)
 
Mosaicking (mosaic.pl)
 
Point Source Extraction (apex*.pl)
 
Basic Concepts
 
Appendix 1: Full List of MOPEX Scripts
 
Appendix 2: Fatal Bit Patterns
 
Appendix 3: Full Lit of MOPEX Modules

Basic Concepts: Tiling

There is no precise limit to the size of a mosaic that MOPEX will produce. However, large datasets may encounter computer memory limitations, so most MOPEX modules employ Tiles to reduce the number of pixels that need to be considered at one time.

Tiles are subsections of the final mosaic image. The size of tiles is set by the user based on the computer memory considerations. It should be small enough so that all input images overlapping one tile can be held in memory without swapping. The software first produces the tiles for each final mosaic image with the tile sizes specified by the user. Then all the tiles are glued together into a single mosaic image by the module Mosaic Combine.

The first step is estimation of the tile geometry. The interpolated and co-added images are defined on the same Fiducial Image Frame (FIF). The suggested tile sizes in the x- and y- directions TILE_XSIZ and TILE_YSIZ are namelist parameters. If both FIF dimensions NAXIS1 and NAXIS2 are smaller than TILE_XSIZ and TILE_YSIZ, correspondingly, there is only one output co-added image in this case.


An example of tiling a FIF. Suppose NAXIS1 = 4450 ad NAXIS2 = 1024. The suggested sizes TILE_XSIZ = TILE_YSIZ = 500. In this case there are 2 tiles in the y-directions of the same size TileY = 512 and 9 tiles in the x-direction, 8 of the size TileX = 494 and the last tile of the size LastTileX = 498.


If the FIF dimensions are greater than TILE_XSIZ or TILE_YSIZ, then the coadded image is tiled. Given the dimensions of the FIF (NAXIS1 and NAXIS2) the program finds the number of the tiles, their sizes and relative positions (offsets) with respect to the FIF. The simple algorithm below assures that all the tiles are of almost the same size regardless of what the FIF size is (see Figure 1).

N_X = NAXIS1 / (TILE_XSIZ);

If (NAXIS1 % TILE_XSIZ > TILE_XSIZ/2)     N_X = N_X +1;

TileX = NAXIS1 / N_X;   LastTileX = TileX + (NAXIS1%N_X)

The % (modulo) operator yields the remainder from the division of the first argument by the second. The calculations in the y-direction are identical. The tiles sizes and offsets from the origin of the FIF along with their names are written in output IPAC table coadd_tiles.tbl.



SPITZER HOME || SPITZER SCIENCE || INSTRUMENTS || SCIENCE USER SUPPORT || SEARCH

help@spitzer.caltech.edu

This file was last modified on Tue Jun 10 17:37:42 PDT 2008.

California Institute of Technology Jet Propulsion Laboratory NASA