Image Filtering
Learn
more about Filtering in SPRING
The Filtering techniques over digital images in SPRING can be accessed at the
main menu: Image - Filtering...
The prerequisites for Image Filtering are:
-
1- having a monochromatic image (in gray levels) and,
-
2- activating this image, not necessarily presenting it on the screen.
NOTE: The steps to perform image filtering are similar for all
types (linear, non-linear or radar). Depending on the type chosen the interface
changes, adjusting to your choice. The procedure for editing and executing a
mask of a new filter and is similar, but you will have to edit the mask before.
Execute
Filtering:
-
activate an Image Infolayer on the "Control Panel" (
);
-
click on Image on the main menu and choose Filtering...;
-
choose the Type (Linear, Non Linear or Radar).
For each type chosen, a group of filters is presented Check the table below to
see the available options;
-
type in the Name
(output layer) on the text box, the IL name to be generated (maximum of 32
characters);
-
in case you want to define an area smaller than the project's, click on Bounding
Box... (see more on bounding box);
-
define the Number of iterations, to define how many times the
same filter will be applied on the same image;
-
click on Apply .
Types of Filters |
Options |
Masks |
Linear |
Low-Mid Pass |
3x3, 5x5 or 7x7 |
Edge Directional |
NW, W, SW, N, S, NE, E or SE |
Edge Non-Directional |
Low, Average or High
|
TM Enhancement |
Predefined by the system (4x4) |
Masks |
Masks editor |
Non Linear |
Edge Detection
|
Sobel or Roberts |
Morphologic Dilation
|
Mtot, Mx, M+, M-, M, Md or Me |
Morphologic Median |
Morphologic Erosion
|
Mask |
Elements Editor |
Radar
|
Lee
|
Define Type, Window (3x3, 5x5, 7x7, 9x9), Image and Number of
Looks.
|
Kuan |
Frost
|
Define Type, Window (3x3, 5x5, 7x7, 9x9), Image, Number of Looks and
Correlation Coefficient. |
Note: the sequence of operations for executing the Filtering is
similar if the mask is defined by the user. You need to select a previously
created mask and execute it on the chosen image. Check next how to define a
mask.
Masks Editor
In SPRING the user can create his own filters. A mask editor allows to define
them for Linear and Non linear filter types.
-
For linear filters the user has the option to create mask with
several dimensions using the values 1,3,5 and 7, for instance,
1x3, 3x3, 7x5
etc...
-
For non-linear filters the user has the options Dilation,
Median and Erosion, using a 3x3 mask for the weights
for the structuring element.
Creating a linear filter
mask:
-
click on Image - Filtering...
on the main menu;
-
click on Linear
to choose the type;
-
choose the option Masks at Linear
Filters. Notice that a text box is presented so you can inform the
mask name or you can use the button Select...
;
-
click on Select... to create a new mask or
choose an existing one from the list. The window "Filtering"
is presented;
-
after creating or updating a mask, you will be able to Execute
Filtering (the last four steps).
Creating a non linear filter
mask:
-
click on Image - Filtering...
on the main menu;
-
click on Non Linear
to choose the type;
-
choose the option Masks at Non Linear
Filters. Notice that a text box is presented so you can inform the
mask name or you can use the button Select...
;
-
click on Select... to create a new mask. The
window "Structuring Element"
is presented;
-
after creating or updating a mask, you will be able to Execute
Filtering (the last four steps).
Filtering Radar Images
When you select the Radar type, the interface is changed and
the parameters are predefined (you will not be able to create a mask).
Executing Radar Filtering;
-
click on Radar
to choose the type;
-
choose the Speckle Noise Filters: Lee, Kuan or Frost;
-
define the parameters in Type (Fixed or Adaptative), Window
(3x3, 5x4, 7x7 or 9x9), Image (Amplitude or Intensity) and Number of
Looks;
-
define a name for the Image Output;
-
change the Bounding Box...
if you wish;
-
click on Apply .
NOTE: The number of Looks represents how many "looks"
the image was generated and this information is usually stored on the header of
the image (see "Multi-Look Processing").
For the Frost filter you should define the correlation Coefficient. (LAG =1,
indicates autocorrelation on a pixel neighborhood). This value can be
calculated at the SPRING statistical module.
See also:
About RADAR Images
How to obtain information about Filtering techniques
Main Menu - Image
|