Index of /806atari/701-800/721
Name Last modified Size Description
![[DIR]](/icons/back.gif)
Parent Directory 24-Sep-2006 04:39 -
![[TXT]](/icons/text.gif)
-SKELET.HLP 19-Jan-1988 14:33 1k
ADD.HLP 19-Jan-1988 14:33 1k
AIM.INF 19-Jan-1988 14:33 1k
AIM2_3.PRG 19-Jan-1988 14:33 139k
AIMBW.RSC 19-Jan-1988 14:33 21k
AIMCOLOR.RSC 19-Jan-1988 14:33 21k
AND.HLP 19-Jan-1988 14:33 1k
ASBEST.IM 19-Jan-1988 14:33 64k
BCOPY.HLP 19-Jan-1988 14:34 1k
BDISP.HLP 19-Jan-1988 14:34 1k
BNOISE.IM 19-Jan-1988 14:34 64k
CADD.HLP 19-Jan-1988 14:34 1k
CDIV.HLP 19-Jan-1988 14:34 1k
CERMET.IM 19-Jan-1988 14:34 64k
CMDFILE.HLP 19-Jan-1988 14:34 1k
CMDWIND.HLP 19-Jan-1988 14:36 1k
CMUL.HLP 19-Jan-1988 14:36 1k
CONTOUR.HLP 19-Jan-1988 14:36 1k
CONTROL.ACC 08-Oct-1986 12:49 20k
COPY.HLP 19-Jan-1988 14:36 1k
CREADF.HLP 19-Jan-1988 14:36 1k
DEMO.AIM 19-Jan-1988 14:36 1k
DEMO1.AIM 19-Jan-1988 14:36 1k
DEMO2.AIM 19-Jan-1988 14:36 1k
DEMO3.AIM 19-Jan-1988 14:36 1k
DEMO4.AIM 19-Jan-1988 14:36 1k
DEMO6.AIM 19-Jan-1988 14:37 1k
DEMO7.AIM 19-Jan-1988 14:37 1k
DILATION.HLP 19-Jan-1988 14:37 1k
DIV.HLP 19-Jan-1988 14:37 1k
DOFF.HLP 19-Jan-1988 14:40 1k
DON.HLP 19-Jan-1988 14:40 1k
ENDPIX.HLP 19-Jan-1988 14:40 1k
EQUAL.HLP 19-Jan-1988 14:40 1k
EROSION.HLP 19-Jan-1988 14:40 1k
EXIT.HLP 19-Jan-1988 14:40 1k
EXOR.HLP 19-Jan-1988 14:40 1k
FILTER.HLP 19-Jan-1988 14:40 1k
GDISP.HLP 19-Jan-1988 14:40 1k
GRADX.HLP 19-Jan-1988 14:41 1k
GRADY.HLP 19-Jan-1988 14:41 1k
HARA.HLP 19-Jan-1988 14:41 1k
HAVE.HLP 19-Jan-1988 14:41 1k
HELP.HLP 19-Jan-1988 14:41 1k
HIST.HLP 19-Jan-1988 14:41 1k
INVERT.HLP 19-Jan-1988 14:42 1k
KETEL.IM 19-Jan-1988 14:42 64k
LABEL.HLP 19-Jan-1988 14:42 1k
LAPLACE.HLP 19-Jan-1988 14:42 1k
LIFE.HLP 19-Jan-1988 14:42 1k
LINKPIX.HLP 19-Jan-1988 14:42 1k
LMAX.HLP 19-Jan-1988 14:42 1k
LMIN.HLP 19-Jan-1988 14:42 1k
LOG1.AIM 19-Jan-1988 14:43 0k
LOGGING.HLP 19-Jan-1988 14:43 1k
MAAN.IM 19-Jan-1988 14:43 64k
MAJORITY.HLP 19-Jan-1988 14:43 1k
MAXIMUM.HLP 19-Jan-1988 14:43 1k
MEDI.HLP 19-Jan-1988 14:43 1k
MINDEV.HLP 19-Jan-1988 14:43 1k
MINIMUM.HLP 19-Jan-1988 14:44 1k
MUL.HLP 19-Jan-1988 14:44 1k
OR.HLP 19-Jan-1988 14:44 1k
PROPAG.HLP 19-Jan-1988 14:44 1k
![[ ]](/icons/unknown.gif)
READ.ME 08-Oct-1986 12:56 1k
READF.HLP 19-Jan-1988 14:44 1k
REMOVE.HLP 19-Jan-1988 14:45 1k
RESET.HLP 19-Jan-1988 14:45 1k
SET.HLP 19-Jan-1988 14:45 1k
SETUP.HLP 19-Jan-1988 14:45 1k
SINPIX.HLP 19-Jan-1988 14:45 1k
SKELET.HLP 19-Jan-1988 14:46 1k
SNAPSHOT.ACC 19-Jan-1988 14:46 7k
SNAPSHOT.NEO 19-Jan-1988 14:46 13k
SUB.HLP 19-Jan-1988 14:46 1k
THRESH.HLP 19-Jan-1988 14:46 1k
TIME.INF 08-Oct-1986 12:56 1k
TOFF.HLP 19-Jan-1988 14:46 1k
TON.HLP 19-Jan-1988 14:47 1k
TRUI.IM 19-Jan-1988 14:47 64k
TRUI.PI1 19-Nov-1985 17:12 12k
TUD.IM 19-Jan-1988 14:51 64k
UNIF.HLP 19-Jan-1988 14:52 1k
VERTICES.HLP 19-Jan-1988 14:52 1k
WRITF.HLP 19-Jan-1988 14:52 1k
AIM : ATARI IMAGE MANAGER
The image processing program AIM 2.3 for the ATARI ST has been developed
by Frans Groen and Robert de Vries of the Delft University of Technology.
Frans is a scientific staff member of of the section "Pattern Recogni
tion" of the vakgroep Signal / System Techniques (ST), Robert is a
student of another section of the same vakgroep.
AIM runs in high-resolution and low-resolution mode. The medium-resolu
tion mode is not supported.
AIM has been written in the C-language using the Lattice C-compiler. The
purpose of the program is to experiment with digital image processing.
User interface
The user interface of AIM consists of:
-GEM interface. Operations and parameters can be selected by means of
dialog boxes and drop-down menus. A choice from a menu results in a
command which is passed to the command interpreter. Image displays can be
moved and sized.
-Command interpreter. Sequences of commands can be put together in a
command file. When the logging is on, commands which are entered are also
stored in a log command file. A command file may be called within another
command file up to 16 levels deep.
Images in AIM
Four image memories are available in AIM 2.3. The image memories are 8
bit wide and have a maximum size of 65536 pixels. Standard aim images
have a the size of 256 x 256 pixels. These image memories are displayed
at the ATARI screen. The windows are designated with 'A', 'B', 'C' and
'D'. The position of the windows is obtained from a setup file. Image
windows can be moved and sized.
A colour image consists of 4 memories with a maximum of 16384 pixels
respectively for the intensity, red, green and blue sub-images. This
maximum size is the result from the current 1M byte memory restriction of
the Atari ST. When the memory increases, larger colour images could be
supported. Standard aim colour images have a size of 4 times 128 x 128
pixels. They occupy the same area as an grey-value image of size 256 x
256. Most of the (colour) image operations are applied to all sub-images
of a colour image. Some image operations (gradx, grady, equal ) are
performed on the intensity sub-image of a colour image. After such an
operation the red, green and blue colour values are updated according to
the new intensity value.
The 8 bitplanes of the grey-value images can also be used as 8 binary
images. Binary and Cellular Logic operations can be performed on these
bitplanes. The bitplanes are numbered from 1 to 8.
When an image is read from disk all 8 bitplanes are read simultaneously.
The same holds when an image is written to disk.
An image is displayed in the grey-value mode when the extension is .IM
and in color mode when the extension is .COL. The display of color images
is only supported in the low-resolution mode.
Also DEGAS and Neochrome images can be read or written. With the read
command these pictures are converted to 320 x 200 grey-value images. With
the cread command they are converted to colour images of size 160 x 100.
When the extension is .PI1 or .PI3 a DEGAS file is read or written, when
the extension is .NEO a neochrome file is read or written. Examples are
given in @democ4.
Great care should be taken by storing images in degas or neochrome
format. In these formats a limited number of grey values or colours is
used (palette), so a drastic reduction of information takes place, which
limits the possibility of further image processing. Images are stored in
full grey or colour resolution in the AIM format.
In the high resolution mode a standard grey-value image is represented by
256 x 256 black and white dots. (there are only 2 grey values possible:
white and black). A sigma-delta technique is used to obtain a local
density of the dots representing as good as possible the grey-values in
the image.
In a binary display 0 gives a black dot and 1 a white dot.
In the low resolution mode a standard grey-value image is represented by
128 x 128 pixels in 8 grey-values. Again a sigma-delta technique is used
to represent as good as possible the original 256 grey values. As the
standard images have a size of 256 x 256 pixels, in the low-resolution
mode only 1 pixel out of each 2x2 pixel neighbourhood is displayed. In
the default binary low-resolution mode (don 1) a black dot is displayed
when all pixels of a 2x2 neighbourhood are '0' and a white dot is
displayed when all pixels in a 2x2 neighbourhood are '1'. In all other
cases a grey dot is displayed.
A standard color image is displayed in 128 x 128 pixels with 16 different
colors. An optimum palette is chosen for the currently displayed image.
So the palettes of other images are no longer correct. The palette of the
image which is on top is displayed. Clicking on another window brings it
on top with the correct palette.
Standard grey-value images on disk (extension .IM) consist of 256 lines
of 256 pixels (unsigned characters) of 1 byte. So an image consists of
65536 bytes in total. The sequence is line by line from top left to
bottom right.
Standard color images on disk (extension .COL) consist of a sequence of 4
sub-images of 128 lines of 128 pixels (unsigned characters) of 1 byte.
The sub-images are intensity, red, green and blue respectively.
When an image does not have a standard format a header is needed
(extension .HD), describing the image width and height. The header used
is the header of the Centre for Image Processing Delft and is given in
the Appendix. When an image is written in the AIM format always a header
is written.
Command interpreter
It is not necessary to give the complete command name. The smallest
number of letters which gives no ambiguity is sufficient.
It is possible to edit your commands in a simple way. The delete- or
backspace-key delete the last character. CONTROL-X deletes the whole
line.
A sequence of commands which perform a certain operation may be grouped
to a command file and stored on disk. The extension of command files is
.AIM. When the logging is on all the entered commands are stored. When
the logging is switched off a log command file is written to disc.
Command files are executed by typing @name (without the extension). A new
command file can be called from a command file and may be nested up to 16
levels deep. So typing @command-file-name instead of a command executes
the command-file. ESC stops the execution of a commandfile and asks the
user to go on or not. Default values are used when a parameter is not
given. When an erroneous value of a parameter is given, the command
interpreter asks for a correct value. If in this case a return is given,
the default value is used. This gives the possibility to ask for a
parameter value when a command-file is executed, instead of using a fixed
value. In the commandfile for instance a question mark (?) can be used
instead of the parameter value to obtain this feature.
A question mark (?) instead of a command gives a list of all available
commands.
A star (*) indicates that the remainder of the commandline consists of
comments.
GEM-INTERFACE
Operations marked with a star (*) are only available in the command
interpreter.
Dialog boxes
Parameters are entered in the dialog boxes by first pushing the ESC
button to delete the default value and to go to the first position. Only
at the first position minus signs are allowed.
There are two special dialog-boxes. One is for the threshold command. The
default threshold value is computed each time a new input-image is
selected.
The other special box is for the palette. If a color is chosen the color
coefficients are shown in the editable text line underneath. If the
coefficients are changed and any other color-box is clicked the effect of
the changes are visible. The three coefficients represent the red, green
and blue, and may vary from 0 to 7. Higher values are cast to 7.
OPERATIONS IN AIM
FILE OPERATIONS
readf[ile]: reads an image from disk.
The first command parameter is the image name, the second optional
parameter is the image memory (A,B,C,D). Default is image memory A. The
extension of an grey-value image file is .IM. Available standard images
are : TRUI.IM, CERMET.IM, TUD.IM and ASBEST.IM. When no extension is
given .IM is assumed.
(When the extension is .COL a colour image is read in).
When no header file is present a standard image size is assumed,
otherwise the description from the header file is used.
When the extension is .PI1 or .PI3 a DEGAS file is converted to a grey-
value image of 320 x 200 pixels. When the extension is .NEO a neochrome
file is converted to a grey-value image of 320 x 200 pixels.
command : readf image-file, image
default : readf trui , A
range : - , <A-D>
creadf[ile]: reads a colour image from disk.
The first command parameter is the image name, the second optional
parameter is the image memory (A,B,C,D). Default is image memory A.
The extension of a color image is .COL. Available color images are :
MANDRIL.COL, MAID.COL, GIRL.COL and ARLES.COL.
When no extension is given .COL is assumed.
(When the extension is .IM a grey-value image is read in).
When no header file is present a standard image size is assumed.
otherwise the description from the header file is used.
When the extension is .PI1 a DEGAS file is converted to a colour image of
160 x 100 pixels. When the extension is .NEO a neochrome file is
converted to a colour image of 160 x 100 pixels.
command : creadf image-file, image
default : creadf mandril , A
range : - , <A-D>
writf[ile]: writes an image to disk.
The first command parameter is the image memory (A,B,C,D). The second
parameter is the name of the image-file. When no extension is given .IM
is assumed. Besides the image file also a header file is written
(extension .HD). The header structure is given in the Appendix.
When the extension is .PI1 or .PI3 a DEGAS file is written. When the
extension is .NEO a neochrome file is written.
command : writf image, image-file
default : writf A , dummy
range : <A-D>, <anything>
load setup
This command loads a setup file. The setup file describes the path of the
image files (read/write), the commandfiles, the helpfiles, the aspect and
align-flags and the position of the windows.
save setup
This command saves the current settings (paths and window positions) in a
setup files.
Macro : selects and starts a commandfile.
@filename is a command file. An @ starts the command file. Command files
may be nested up to 16 levels. An example is given in @demo. The
extension of command files is .AIM. These extensions are added to the
command file name by AIM.
When infinite = -1 the commandfile is infinitely repeated.
command : @name.AIM,infinite
default : - , 0
range : - , <-1,0>
Logging
When the logging is on all entered commands are stored. When the logging
is switched off, the stored commands are written in a log file. In this
way automatically command files can be created.
exit : ends the program.
This action can also be obtained by typing Control-C or by clicking the
close-box of the command-window.
command : exit
ARITHMETIC GREY-VALUE OPERATIONS
copy : copies an image.
The command parameters are the input image and the output image
(A,B,C,D).
command : copy input-image , output-image
default : copy A , B
range : <A-D> , <A-D>
add : adds two images.
The first command parameter is the input-image. The second parameter is
the output-image to which the input-image is added.
output-image = (input-image + output-image )/2
command : add input-image , output-image
default : add A , B
range : <A-D> , <A-D>
sub : subtracts two images.
The first parameter is the input-image. The second parameter is the
output-image from which the input-image is subtracted. The third parame
ter is a constant which can be added to the image. As the image cannot
become negative, the resulting value is clipped below 0 and above 255.
output-image = (output-image - input-image ) + constant
command : sub input-image , output-image , constant
default : sub A , B , 0
range : <A-D> , <A-D> , <-255,255>
mul : multiplies two images.
The output-image (second parameter) is multiplied by the input-image
(first parameter).
output-image = (input-image * output-image )/256
command : mul input-image , output-image
default : mul A , B
range : <A-D> , <A-D>
div : divides two images.
The output-image (second parameter) is divided by the input-image (first
parameter).
output-image = (output-image / input-image )
command : div input-image , output-image
default : div A , B
range : <A-D> , <A-D>
(*) minimum : minimum of two images.
The output-image (second parameter) is the minimum of the output-image
and the input-image (first parameter).
output-image = minimum (input-image, output-image )
command : minimum input-image , output-image
default : minimum A , B
range : <A-D> , <A-D>
(*) maximum : maximum of two images.
The output-image (second parameter) is the maximum of the output-image
and the input-image (first parameter).
output-image = maximum (input-image, output-image )
command : maximum input-image , output-image
default : maximum A , B
range : <A-D> , <A-D>
(*) mindev : take input value with minimum deviation.
The output pixel value is that value of one of the two input images, that
differs the least from the output pixel value.
if |output value - input1 value| < |output value - input2 value|
output = input1
else
output = input2
command : mindev input1-image , input2-image, output-image
default : mindev A , B , C
range : <A-D> , <A-D> , <A-D>
cadd : adds a constant to an image.
The first command parameter is the image. The second parameter is the
constant which is added to the image.
command : cadd image , constant
default : cadd A , 0
range : <A-D> , <-255,255>
cmul : multiplies an image by a constant.
The first command parameter is the image. The second parameter is the
multiplication constant.
command : cmul image , constant
default : cmul A , 1
range : <A-D> , <-255,255>
cdiv : divides an image by a constant.
The first command parameter is the image. The second parameter is the
division constant.
command : cdiv image , constant
default : cdiv A , 1
range : <A-D> , <-255,255>
GREY-VALUE FILTERS
A pixel value in the output image is determined by the value of the pixel
in the input image together with the pixel values in a local neighbour
hood around the pixel. Both linear and non-linear operations can be
applied to such a neighbourhood.
LINEAR
A linear filter can be realized by multiplying a pixel and its
neighbourhood by coefficients. The result is the sum of these multiplica
tions and is called a convolution. Examples of linear filters are the
'laplace' and 'gradient' filters.
In AIM the size of linear filters is limited to a 3x3 neighbourhood.
'Filter' is a rotation-symmetrical 3x3 filter of which the coefficients
can be specified by the user. In the the 'laplace' and 'gradient' filters
the coefficients are fixed.
The coefficients are :
-1 0 -1 -1 -2 -1 0 -1 0 cd chv cd
-2 0 -2 0 0 0 -1 5 -1 chv cc chv
-1 0 -1 -1 -2 -1 0 -1 0 cd chv cd
x-gradient y-gradient laplace filter
laplace : linear filter for image enhancement.
The first command parameter is the input-image and the second parameter
is the output-image. Defaults are: A for the input-image and B for the
output-image.
command : laplace input-image , output-image
default : laplace A , B
range : <A-D> , <A-D>
gradx : horizontal gradient filter.
The first parameter is the input-image (default A) and the second
parameter is the output-image (default B). The output-image may be
multiplied by a factor (multiply/1024) and a constant (add) may be added
to the result.
output-image = gradx(input-image ) * (multiply / 1024) + add
command : gradx input-image , output-image , multiply, add
default : gradx A , B , 1024 , 128
range : <A-D> , <A-D> ,<1-10000>,<-255-255>
grady : vertical gradient filter.
The first parameter is the input-image (default A) and the second
parameter is the output-image (default B). The output-image may be
multiplied by a factor (multiply/1024) and a constant (add) may be added
to the result.
output-image = grady(input-image ) * (multiply / 1024) + add
command : grady input-image , output-image , multiply, add
default : grady A , B , 1024 , 128
range : <A-D> , <A-D> ,<1-10000>,<-255-255>
filter : 3x3 rotation-symmetrical filter.
This filter performs a general 3x3 rotation-symmetrical convolution.
Parameters are the input-image, the output-image and the 3 different
coefficients in a 3x3 neighbourhood. The coefficient of the central pixel
is cc, the coefficients of the horizontal and vertical neighbours are chv
and the coefficients of the diagonal neighbours are cd. The result may be
multiplied by a factor (multiply / 1024) and a constant (add) may be
added to the result.
output-image = filter(input-image ) * (multiply/1024) + add
command : filter input-image, output-image, cd, chv, cc, multiply, add
default : filter A , B , 0 , -1 , 4 , 2048 , 0
range : <A-D> , <A-D> , <-/+1000> , <10000> ,<-+255>
unif : uniform filter.
This filter determines for each pixel the average of a local neighbour
hood in the input-image. This average becomes the pixel value in the
output-image.
The first parameter is the input-image, the second parameter is the
output-image. The third parameter is the size of the local neighbourhood.
This size is an odd value between 1 and 63.
command : unif input-image ,output-image , neighbourhood size
default : unif A , B , 3
range : <A-D> , <A-D> , <1,63>
NON-LINEAR
Non-linear filters perform image operations which are impossible with
linear filters. Noise can be suppressed with a non-linear filter without
deteriorating edges in an image. Only the median filter, the minimum and
the maximum filter are implemented in AIM.
medi : median filter.
The median filter determines for each pixel the median (the middle) value
of a local (3x3) neighbourhood in the input-image. This median value
becomes the pixel value in the output-image.
This filter suppresses noise.
The first parameter is the input-image, the second parameter is the
output-image.
command : medi input-image ,output-image
default : medi A , B
range : <A-D> , <A-D>
lmin : local minimum filter.
The minimum filter determines for each pixel the minimum of a local
neighbourhood in the input-image. This minimum becomes the pixel value in
the output-image. This operation results in an extension of the dark
parts in an image. This operation is also called grey-value erosion.
The first parameter is the input-image, the second parameter is the
output-image. The third parameter is the size of the local neighbourhood.
This size is an odd value between 1 and 63. The operation is illustrated
in @demo5 and @demo7.
command : lmin input-image ,output-image , neighbourhood size
default : lmin A , B , 3
range : <A-D> , <A-D> , <1,63>
lmax : local maximum filter.
The maximum filter determines for each pixel the maximum of a local
neighbourhood in the input-image. This maximum becomes the pixel value in
the output-image. This operation results in an extension of the bright
parts of an image. This operation is also called grey-value dilation.
The first parameter is the input-image, the second parameter is the
output-image. The third parameter is the size of the local neighbourhood.
This size is an odd value between 1 and 63.
command : lmax input-image ,output-image , neighbourhood size.
default : lmax A , B , 3
range : <A-D> , <A-D> , <1,63>
(*)hara : non-linear edge filter.
The filter hara performs a variant (Lucas van Vliet) of the 3x3 Lee-
Haralick edge filter (Patt. Rec. Conf., Paris '86).
This filter is based upon the weighted difference between the original
image and the local minimum and local maximum filtered images.
The first command parameter is the input-image and the second parameter
is the output-image. Defaults are: A for the input-image and B for the
output-image.
command : hara input-image , output-image
default : hara A , B
range : <A-D> , <A-D>
(*)have : non-linear laplace like filter.
The filter have performs a non-linear laplace like-like operation
inspired by the 3x3 Lee- Haralick edge filter (Patt. Rec. Conf., Paris
'86).
Adding the have result to the original gives sharpening as in the linear
laplace filter.
This filter is based upon the weighted difference between the original
image and the local minimum and local maximum filtered images.
The first command parameter is the input-image and the second parameter
is the output-image. Defaults are: A for the input-image and B for the
output-image.
command : have input-image , output-image
default : have A , B
range : <A-D> , <A-D>
equal(ization) : histogram equalization.
This operation reassigns the grey-values in an image and results in an
"optimum" use of the available grey-values. The grey-value histogram of
the output image is as flat as possible (given the necessary requantiza
tion). An example is given in @demo1.
command : equal input-image , output-image
default : equal A , B
range : <A-D> , <A-D>
thresh[olding]: transforms a grey-value image in a binary image.
The binary image is obtained by thresholding the input image. After
thresholding binary image operations may be applied. Pixels with a grey-
value above the threshold get the binary value '1' and thus become white.
Pixels with a grey-value below the threshold get the binary value '0' and
become black. The default value of the threshold is obtained from the
grey-value histogram with the method of Ridler & Calvert.
command : thresh input-image , output-image , bitplane, threshold
default : thresh A , B , 1 , automatic
range : <A-D> , <A-D> , <1,8> , <0,255>
CELLULAR LOGIC (BINARY) OPERATIONS
These operations can be applied to the 8 bitplanes of an image. The
bitplanes are numbered from 1 to 8. A black pixel has the value '0', a
white pixel has the value '1'.
The edge value of the image can be set to '0' or '1' in all operations.
The default value of the image edge is black (0). The given number of
iterations is a maximum. The iterations stop when no changes in the pixel
values occur anymore.
The number of iterations is shown in that case.
erosion : erodes white objects.
One layer of the white objects is peeled off in each iteration. White
objects which are smaller than the number of iterations * 2 will
disappear. The connectivity determines which white pixels belong to a
layer to be peeled off. When the connectivity is 4 this layer consists of
those white pixels that have a horizontal or vertical black neighbour.
When the connectivity is 8 a layer consists of those white pixels that
have a horizontal, vertical or diagonal black neighbour. When the
connectivity is 6 for the odd iterations a 4-connectivity is used and for
the even iterations an 8-connectivity is used. The erosion is illustrated
in @demo6.
command : erosion #iterations , bitplane, connectivity, edge ,image
default : erosion 1 , 1 , 6 , 0 , B
range : <1,512> , <1,8> , <4,6,8> , <0,1>,<A-D>
dilation : extends white objects.
One layer is added to the white objects in each iteration. Black parts
such as holes which are smaller than the number of iterations * 2 will
disappear. The connectivity determines which black pixels belong to a
layer to be added. When the connectivity is 4 this layer consists of
those white pixels that have a horizontal or vertical white neighbour.
When the connectivity is 8 a layer consists of those black pixels that
have a horizontal, vertical or diagonal white neighbour. When the
connectivity is 6 for the odd iterations a 4-connectivity is used and for
the even iterations an 8-connectivity is used.
command : dilation #iterations, bitplane, connectivity, edge, image
default : dilation 1 , 1 , 6 , 0 , B
range : <1,512> , <1,8> , <4,6,8> ,<0,1>, <A-D>
propag[ation]: dilates a bitplane under a mask.
The propagation strongly resembles the dilation. However, the black
pixels are only added when they are white (1) in the mask bitplane. The
propagation can be used to find the original shapes of objects which
remained after a number of erosion iterations. This application is also
shown in @demo6.
The first parameter is the number of iterations. The second parameter is
the bitplane of which the dilation takes place (seed bitplane). The third
parameter is the mask bitplane. The other parameters are again the
connectivity and the image edge value. When the edge value is 0 only the
seed bitplane is propagated. When the edge value is 1 only the edge is
propagated and when the edge value is 2 both the edge and the seed
bitplane are propagated.
command : propag #iterations,bitplane,maskbitplane,connectiv, edge,image
default : propag 1 , 1 , 2 , 6 , 0 , A
range : <1,512> , <1,8> , <1,8> , <4,6,8> ,<0,2>,<A-D>
skelet : skeleton with endpixels.
The skeleton is an 1 pixel thick line in the middle of white objects. The
skeleton operation resembles the erosion, but in this case pixels of a
layer are only deleted, when their deletion does not change the topology
of the image. (They are not break pixels). An additional restriction is
that a pixel may not be deleted at the end of a skeleton(line). The
number of iterations must be sufficiently large to find the complete
skeleton (the biggest diameter of the white objects divided by 2).
The operation stops when the number of iterations is reached or when no
pixels change anymore. An example is given in @demo3.
command : skelet #iterations , bitplane, edge value, image
default : skelet 100 , 1 , 0 , B
range : <1,512> , <1,8> , <0,1> , <A-D>
-skelet : skeleton without endpixels
The skeleton without endpixels is the same as the previous skeleton
operation, but endpixels may be deleted in this case. Solid objects will
result in single pixels and holes in objects will result in rings. This
is illustrated by the skeleton of the background (exo-skeleton) in
@demo4.
command : -skelet #iterations , bitplane, edge value, image
default : -skelet 100 , 1 , 0 , B
range : <1,512> , <1,8> , <0,1> , <A-D>
contour : determines the contours of objects.
The result is the edges of white objects. This means all white pixels
with one or more black neighbours.
command : contour bitplane, edge value, image
default : contour 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
majority : performs a majority voting.
The central pixel of a 3x3 neighbourhood becomes black or white depending
upon the majority of the neighbourhood.
command : majority bitplane, edge value, image
default : majority 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
remove[noise]: removes pepper and salt noise.
A white pixel in a complete black 3x3 neighbourhood becomes black. A
black pixel in a complete white 3x3 neighbourhood becomes white.
command : remove bitplane, edge value, image
default : remove 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
life : game of life
The well-known game of life is also a cellular logic operation !
command : life #iterations , bitplane, edge value, image
default : life 1 , 1 , 0 , B
range : <1,512> , <1,8> , <0,1> , <A-D>
SPECIAL LINEFIGURE PIXELS
sinpix[els] : determines isolated white pixels
These white pixels have no white neighbour in a 3x3 neighbourhood.
command : sinpix bitplane, edge value, image
default : sinpix 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
endpix[els]: determines endpixels of a linefigure.
These white pixels have only one white neighbour in a 3x3 neighbourhood.
command : endpix bitplane, edge value, image
default : endpix 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
linkpix[els]: determines white pixels with two white neighbours.
These white pixels have two white neighbours in a 3x3 neighbourhood.
command : linkpix bitplane, edge value, image
default : linkpix 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
vertices: determines the vertices in a linefigure.
These white pixels have more than two white neighbours in a 3x3
neighbourhood.
command : vertices bitplane, edge value, image
default : vertices 1 , 0 , B
range : <1,8> , <0,1> , <A-D>
BIT OPERATIONS
bcopy : copies a bitplane.
The command parameters are the input-bitplane and the output-bitplane.
command : bcopy input-bitplane,output-bitplane,input-image,output-image
default : bcopy 1 , 3 , B , B
range : <1,8> , <1,8> , <A-D> , <A-D>
and : AND-function of two bitplanes.
The command parameters are the input-bitplanes of which the AND- function
is taken and the output-bitplane.
command : and bitplane in 1 , bitplane in 2 , bitplane out, image
default : and 1 , 2 , 2 , B
range : <1,8> , <1,8> , <1,8> , <A-D>
or : OR-function of two bitplanes
The command parameters are the input-bitplanes of which the OR- function
is taken and the output-bitplane.
command : or bitplane in 1 , bitplane in 2 , bitplane out, image
default : or 1 , 2 , 2 , B
range : <1,8> , <1,8> , <1,8> , <A-D>
exor : EXOR-function of two bitplanes
The command parameters are the input-bitplanes of which the EXOR-function
is taken and the output-bitplane.
command : exor bitplane in 1 , bitplane in 2 , bitplane out, image
default : exor 1 , 2 , 2 , B
range : <1,8> , <1,8> , <1,8> , <A-D>
invert : inverts a bitplane
The first command parameter is the bitplane. The second command parameter
is the image.
command : invert bitplane, image
default : invert 1 , B
range : <1,8> , <A-D>
set : sets a bitplane, so makes it white (1).
The first command parameter is the bitplane. The second command parameter
is the image
command : set bitplane, image
default : set 1 , B
range : <1,8> , <A-D>
reset : resets a bitplane, so makes it black (0).
The first command parameter is the bitplane. The second command parameter
is the image.
command : reset bitplane, image
default : reset 1 , B
range : <1,8> , <A-D>
label : labels connected components.
This command labels the connected components within a binary image. The
maximum number of labels is 255. (In low resolution the best display is
obtained by: don,3 followed by bdisp.)
command : input-image, output-image, input bitplane, connectivity
default : B , A , 1 , 8
range : <A-D> , <A-D> , <1-8> , <4,6,8>
UTILITIES
(*)don : Automatic display on
When the automatic display is on, after each operation the result is
shown.
An optional parameter gives the type of display in the low-resolution
mode. An example is given in @democ5.
0 : display off (same as doff)
1 : When all pixels of a 2x2 neighbourhood in the original bitplane are
'0' a black dot is displayed. When all pixels of a 2x2 neighbourhood in
the bitplane are '1' a white dot is displayed. In all other cases a grey
dot is displayed.
2 : The bitplane is sampled depending upon the GEM window size. When the
first pixel of a sample region is '0' a black dot is displayed. When the
first pixel is '1' a white dot is displayed.
3 : Multiple bitplane display. Bitplanes are sampled depending upon the
GEM windowsize. Bitplane 1 is displayed in red, bitplane 2 in blue and
bitplane 3 in green. When all three lowest bitplanes are '0', the 3
highest bitplanes are shown as grey-value.
4 : Multiple bitplane display. Bitplanes are sampled depending upon the
GEM windowsize. Bitplane 8 is displayed in red, bitplane 7 is displayed
in blue and bitplane 6 is displayed in green.
command : don, display_type
default : don, 1
range : don, < 0,4>
(*)doff : Automatic display off
When the automatic display is off, only an image is displayed after a
gdisp or bdisp command.
command : doff
(*)ton : Timer on
The execution time of each command is displayed when the timer is on.
command : ton
(*)toff : Timer off
Sets timer off.
command : toff
gdisp : Grey-value / color display of an image.
This routine displays a grey-value image as 256x256 black or white dots
in high resolution. In low resolution 128 x 128 pixels of 8 grey-values
are displayed. Images are always displayed with maximum contrast. When
the image is a color image it is displayed in color in the low resolution
mode. Parameter is the image (A to D).
command : gdisp image
default : gdisp A
range : <A-D>
bdisp : Displays one of the 8 bitplanes of an image.
First command parameter is the image (default A). Second parameter is the
bitplane (default 1). In high resolution black or white pixels are
displayed. The type of display in low resolution is determined by the
(optional) don parameter.
command : bdisp image,bitplane
default : bdisp A , 1
range : <A-D>, <1,8>
hist : calculates the histogram of an image.
The histogram is displayed in overlay in the appropriate image window (A
to D). A threshold is calculated from the histogram to segment the image
( for the thresh command). The command parameter is the image (default
A).
command : hist image
default : hist A
range : <A-D>
Palette : Change the palette interactively.
This option is only supported in the low-resolution mode.
Preferences: Change preferences.
When the image windows are sized, the aspects ratio can be preserved or
not.
The images can be aligned to byte boundaries or not.
help : gives help information.
command : help command
APPENDIX : File header structure
/* call the headerfile *.hd, this file contains the structure with
all parameters of the corresponding image file *.im. */
typedef struct
{
short *address; /* image_address in memory */
short format; /* format */
short sizex; /* dimension 1 */
short sizey; /* dimension 2 */
short sizez; /* dimension 3 */
unsigned short commsz; /* size of comment in bytes */
short data[16]; /* user image dependent data*/
}
HEADER;
#define F_WORDS 0 /* format words in the image */
#define F_BYTES 1 /* format bytes in the image */
#define F_FOBIT 2 /* four bits image, packed in 4 pix/word */
#define F_TWBIT 3 /* two bits image, packed in 8 pix/word */
#define F_SBITS 4 /* format single bits in the image */
ACKNOWLEDGEMENT
The following people contributed to AIM:
Frans Groen (TUD) : image processing.
Robert de Vries (TUD) : GEM interface.
Igor Weber (University of Zagreb, YU)
: low-resolution and colour display.
Gert van Antwerpen (CBD) : lmin, lmax, unif, label
Piet Verbeek (TUD) : hara, have filter
Lucas van Vliet and : erosion, dilation, propagation,
Ben Verwer (TUD) skeleton
Jaap Joosten (TUD) : tests and evaluation.
Rijswijk, 1 July 1987.
Please send remarks and comments to :
Frans Groen
Applied Physics Department
Delft University of Technology
Lorentzweg 1
2628 CJ Delft
The Netherlands
tel (31) 15-781416