SAS/SPECTRAVIEW Software User's Guide

Creating and Applying Your Own Filter

To create a filter to suit your own needs, you can customize a supplied filter by modifying its preset values, or you can define a filter using SAS DATA step code.

Customizing a Filter

The following steps create a filter by modifying the preset values of a filter provided with SAS/SPECTRAVIEW:

Load the appropriate data set.
Select [Data], then [Filter data].
Specify the filter template you want to customize by selecting one of the following. You cannot customize the Median filter.

[Blend filter] smooths by averaging without weighting the response value.

[Gaussian filter] smooths by averaging and weights the response value.

[Laplacian filter] sharpens the data.
Select the filter element(s) that you want to have the same value, and adjust the slider under Filter Value to set the value.
Select [Deselect buttons] to release the selected elements.
If needed, adjust more element values.
Select [Apply filter] to apply the customized filter to all response values in the data.
To reuse the customized filter, store it in a SAS data set by selecting [Save filter] and specifying a one- or two-level SAS data set name in the text window.

Creating a Filter with SAS DATA Step Code

Using SAS DATA step code, you can create a SAS data set containing a filter, which can then be loaded into SAS/SPECTRAVIEW. The following restrictions apply:

The data set must contain exactly four numeric variables, named as follows and in this order: FILTER_X, FILTER_Y, FILTER_Z, FILTER.
- FILTER_X, FILTER_Y, and FILTER_Z are the indices of the filter for X, Y, and Z. The valid range for these values is 1, 2, and 3.
- FILTER is the value for a given set of indices. Its range of values is -16 to 16, inclusive.
All values must be integers or are converted to integers when read in. The data set must have at least 27 values, one for each location in the filter. More than 27 values produces a warning message. Incomplete filters are aborted. The data set cannot contain missing values.

The following DATA step code creates a SAS data set named BLEND, which contains a valid filter:

       data blend;
         do filter_x = 1 to 3;
         do filter_y = 1 to 3;
         do filter_z = 1 to 3;
           filter = 1;  output;
         end; end; end;
       run;

Applying a Filter Stored in a Data Set

The following instructions apply a filter stored in a SAS data set:

Load the appropriate data set.
Select [Data], then [Filter data].
Select [Load filter] and enter the data set name in the text window.
Select [Data set filter] to display the filter matrix.
Select [Apply filter] to apply the filter to all response values in the data.

Chapter Contents
Previous
Next
Top of Page

	[Blend filter] smooths by averaging without weighting the response value.
	[Gaussian filter] smooths by averaging and weights the response value.
	[Laplacian filter] sharpens the data.