home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Night Owl 6
/
Night Owl's Shareware - PDSI-006 - Night Owl Corp (1990).iso
/
020a
/
img_fltr.zip
/
FILTERS.DOC
< prev
Wrap
Text File
|
1991-11-25
|
7KB
|
193 lines
The following is a list of all the 3 X 3 image processing filters for use
with a program such as IMPROCES. These kernals can be used in the custom
filter and boost items of the enhance section in the improces program.
This text file is being supplied with two PCX files which have a variable
frequency bar taget and variable frequency sinusoidal targets for use with
the image filters contained within this document. The following kernal
information was taken from the program CONVOLVE. The program CONVOLVE uses
text files to describe the filter function. The image filter information
presented here is from the CONVOLVE program. The limitation of the program
CONVOLVE is that it can only work on 320 X 200 images; the program IMPROCES
has no such limitation. The kernal information from CONVOLVE can now be used
on all VGA/SVGA resolutions using the IMPROCES program. IMPROCES allows the
user to manipulation the 3 X 3 kernal in the images enhancement menu. However
the interger data is manually entered in to the kernal. IMPROCES also has a
boost function linear multiplier which allows some of the low pass filters to be
realized. Thus some of the filter function can not be achieved yet with the
IMPROCES program but it is hoped that shortly this particular feature can be
delivered by this fantastic program.
The following text is taken almost directly from the CONVOLVE program user
documentation.
CONVOLUTION
The concept of convolution is the application of a filter function to an
image to enhance specific aspects of it. Convolution is an area process.
What is meant by an area process is that each pixel in the resultant image is
determined a group of pixels in the imediate area, in the original image.
Specifically this group of pixels is a 3x3 filter, or kernal, that gets
applied to our original image. Each of the filters provided has a certain
mathematical significance. Using the test patterns provided will help the
user see exactly what is happening to the image as each filter is applied.
Also this document contains other filter functions for items which are
contained in the Enhance menu of the IMPROCES program; the filters here can
be used as an alternative to filters provided to yield an additional degree
of freedom.
Blurring Filter
The purpose of this filter is to blurr the image
BLUR: The blurring filter
1 1 1
1 1 1
1 1 1
Gradient Directional Edge Enhancement :
Using Gradient edge detectors allows you to specify a direction for the
edges other than just horizontal or vertical. Diagonal edges can be
specified as well. This is accomplished by using directions to indicate
the exact direction of the edges desired. If a positive slope in the
direction of the filter exists, a light-colored pixel will be placed in the
resultant image. For example if the East Kernal is used, a light-colored
pixel will be placed in the output image if there is a transition from
black to white in the east (left to right) direction of image. Constant
regions will be attenuated while regions of high frequency change will be
accentuated.
GRADEAST: The gradient East filter
-1 1 1
-1 -2 1
-1 1 1
GRADNE: The gradient North East filter
1 1 1
-1 -2 1
-1 -1 1
GRADN: The gradient North filter
1 1 1
1 -2 1
-1 -1 -1
GRADS: The gradient South filter.
-1 -1 -1
1 -2 1
1 1 1
GRADNW: The gradient North West filter.
1 1 1
1 -2 1
-1 -1 -1
GRADIENT SE: The gradient South East filter.
-1 -1 1
-1 -2 1
1 1 1
GRADSW: The gradient South West filter.
1 -1 -1
1 -2 -1
1 1 1
GRADW: The gradient West filter.
1 1 -1
1 -2 -1
1 1 -1
High-Pass Spatial Filters
High-pass filters accentuate the high- frequency details of an image while
leaving the low-frequency content intact. High pass filtering is used
whenever objects with high spacial-frequency content need to be examined.
The higher-frequency portions of an image will be highlighted while the
lowe frequency portions become black. The use of High pass filters may
highlight images at the expense of adding noise to the image. High
frequency in images can be found by looking at edges of objects. Edge
enhancement of an image is possible with the use of High-pass filtering.
HIGHPASS191: High pass spatial filter
-1 -1 -1
-1 9 -1
-1 -1 -1
HIGHPASS150:
0 -1 0
-1 5 -1
0 -1 0
HIGHPASS125:
1 -2 1
-2 5 -2
1 -2 1
Laplacian Edge Enhancement
Laplacian edge enhancement differs from the other enhancement methods since
it is omni- directional. It highlights edges regardless of direction.
Laplacian edge enhancements generate sharper edge definition than do most
other enhancment operation. Additionally, it highlights edges having both
positive and negative brightness slopes. All regions in the image which
illustrate a rapid change will be accentuated and non-varing regions
attenuated.
LAP1:
0 1 0
1 -4 1
0 1 0
LAP2:
-1 -1 -1
-1 8 -1
-1 -1 -1
LAP3:
-1 -1 -1
-1 9 -1
-1 -1 -1
LAP4:
1 -2 1
-2 4 -2
1 -2 1
Low-Pass Spatial Filters
Low-pass filters leave the low-frequency content of an image intact while
attenuating the high frequency content. Low-pass filters are good at
reducing the visual noise contained in an image. Noise is garbage found
in image that does not pertain to image. Low frequency areas in an image is
where the color of the pixels vary slowly or remains constant.
LOWPAS10:
0.1 0.1 0.1
0.1 0.2 0.1
0.1 0.1 0.1
LOWPAS19:
0.1111 0.1111 0.1111
0.1111 0.1111 0.1111
0.1111 0.1111 0.1111
LOWPAS486:
0.0625 0.125 0.0625
0.125 0.25 0.125
0.0625 0.125 0.0625
Shift and Difference Edge Enhancement
As the name implies, these filters enhance image edges by shifting an
image pixel and then substracting the shifted image from the original. The
result of the substraction is a measure of the slope of the brightness trend.
In an area of constant pixel intensity, the substraction will yield a slope
of zero. Zero results in black pixel values. In an area with large changes
in intensity, an edge, for example, the subtraction will yield a large value
for the slope, which will become a light colored pixel. The larger the
difference in intensities, the lighter the pixel.
SDVEDGE: Shift and difference vertical edge
-1 0 0
0 1 0
0 0 0
SDHEDGE: Shift and difference horizontal edge
0 -1 0
0 1 0
0 0 0
SDHVEDGE: Shift and difference vertical and horizontal edges
0 0 0
-1 1 0
0 0 0