SSC: MOPEX Online Manual

Basic Concepts: Background Matching Algorithm

Background matching of the overlapping input images is performed by the MOPEX script overlap.pl. It calculates and applies an additive correction to each image in the input stack in order to bring them to a common background level. This is often the first step in processing Spitzer imaging data, before mosaicking the frames together.

The images interpolated to a common grid can be subtracted pixel-by-pixel in order to match their backgrounds. We assume that the only correction required is a constant offset, εⁿ, i.e. that the following corrections should be applied to the input image Iⁿ

Here Oⁿ is the output corrected image. The metric to be minimized is the combined uncertainty weighted difference between the overlapping parts of each pair of input BCD's:

Here m and n are image indicies; kⁿ is the pixel number in image n. Minimization with respect to εⁿ's

leads to the following set of N_images-1 linear equations:

where

The symbol O^mn respresents the fact that the summation is done over the overlap area of the m-th and n-th images. In Figure 1 the case of 2 overlapping images is shown. The matrix element M¹² is in this case equal to

The actual ranges of indicies symbolized by O^mn are calculated using the input table with the offsets and sizes of the interpolated images.

Since the problem is invariant under any arbitrary global shift δ of all images, one of the shifts can be picked to be the last one ε^Nimages-1. It is set to 0 at first. The additional constraint that will fix the global shift is to have the total shift of all images add up to 0:

The ε's are analyzed before applying the above condition. The outliers amoung the ε's are found. The following parameters are used:

BOTTOM_THRESHOLD,TOP_THRESHOLD: the thresholds for outlier detection in terms of sigma
MIN_IMG_NUM: the minimum number of images required to detect outliers.

If N_images is greater than or equal to MIN_IMG_NUM, then the outlier ε's are excluded from calculating δ, but δ is applied to all the ε's.

Figure 1: The overlap of Image1 and Image2 is an area of four pixels with the following pixel indices: k₁= {36,37,38,39}; k₂ = {0,1,2,3}, given that the first pixel has index = 0 and the x-direction is scanned first.

Contents

Previous: Basic Concepts: Image Interpolation

Next: Basic Concepts: PRF Fitting in MOPEX

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.