# center1d xtools

NAME · SYNOPSIS · ARGUMENTS · DESCRIPTION · BUGS

## NAME

center1d -- One dimensional centering

## SYNOPSIS

```center = center1d (initial, data, npts, width, type, radius, threshold)

real	initial			# Initial guess
real	data[npts]		# Data points
int	npts			# Number of data points
real	width			# Feature width
int	type			# Feature type
real	threshold		# Detection threshold
```

## ARGUMENTS

initial
Initial guess for the center of the feature.
data[npts]
Pixel data vector.
npts
Number of points in the data vector.
width
Width used to define the convolution function. If the width is 1 or less then the nearest minimum or maximum is returned. If the width is greateer than 1 then a minimum with of 3 is used in the algorithm.
type
Type of feature. The feature types are defined in the file <xtools/center1d.h>. Currently the types are emission and absorption features.
threshold
Minimum difference between the maximum and minimum pixel value in the region around the initial guess allowed for detecting a feature. For data which is all positive and the type is emission then the threshold is also used as an absolute cutoff.

## DESCRIPTION

If the width is 1 or less then the nearest minimum or maximum is found. The centering radius is still applied as is the threshold. If the width is greater than 1 then a minimum width of 3 is used in the algorithm.

The one dimensional position of a feature is determined by solving the equation

(1) integral {(I-I0) f(X-XC) dX} = 0

where I is the intensity at position X, I0 is the continuum intensity, X is the pixel coordinate, and XC is the desired feature position. Figure 1 shows the range of pixels used in determining the continuum intensity, the feature threshold, and solving the equation.

```		      Figure 1:  Data Feature Vector
+-----------------------------------------------------------+
-|                               *                           |
S|                              * *                          |
|                              * *                          |
t|                             *   **                        |
|                             *     *                       |
r|                            *       *                      |
|                           *        *                      |
e|                          *          *                     |
|                         *            *                    |
n|                      *                 *              *  *|
|                   *                       *  *  *  *      |
t|             *  *                                          |
|       *  *                                                |
h|    *                                                      |
-| *                                                         |
+---------+-----------------+---------------------+---------+
-B        -A                 0                     A         B

X-XC

A = radius + 0.5 width		B = radius + 1.5 width
```

The range -A to A is used to determine the continuum intensity and the strength of the feature. For absorption features the continuum intensity is the maximum point in this range while for emission features the continuum is set to zero. Admittedly these are not real measures of the continuum but they contain the fewest assumptions and are tolerant of nearby contaminating features. The feature strength is the difference between the maximum and minimum values. If the feature strength is less than the specified detection threshold then a value of INDEF is returned for the feature position.

The range -B to B includes the range of allowed feature positions plus the half-width of the feature. This range is used in solving equation (1). The convolution function f(X-XC) is a sawtooth as shown in figure 2. For absorption features the negative of this function is used.

```		           Figure 2:  f(X-XC)
+-------------------+-------------------+
|                   |     *             |
|                   |   *   *           |
|                   | *       *         |
0  +-*-*-*-*-----------*-----------*-*-*-*-+
|         *       * |                   |
|           *   *   |                   |
|             *     |                   |
+-------+-----------+-----------+-------+
-width/2       0         width/2

X-XC
```

The two figures graphically define the parameter width . Generally it should be set to a value near the actual width of the emission or absorption feature. If the width is too wide then the center will be affected by blending from nearby lines while if it is too narrow the accuracy of the centering is decreased. The parameter radius determines how far from the initial estimate for XC the interative solution of (1) may go. Equation (1) is solved iteratively starting with the initial position. When successive positions agree within 0.1% of a pixel the position is returned. If the position wanders further than radius from the initial guess or outside of the data vector then the procedure returns the value INDEF. If more than 100 iterations are required or the corrections per iteration exceed the minimum correction reached after 3 further iterations then the solution has failed to converge and INDEF is returned. Note that this latter condition may occur if the width is too small in a flat topped profile.

This task uses the one dimensional image interpolation package iminterp in solving equation (1).

## BUGS

Though the algorithm does not fail if the width is made very small the results become unreliable. Therefore a silent limit of 3 is imposed by the algorithm. If there is ever a need to allow smaller widths then the procedure can be changed and the applications relinked.

Source Code · Search Form · STSDAS