You are on page 1of 10

PROCEEDINGS OF SPIE

SPIEDigitalLibrary.org/conference-proceedings-of-spie

Special-purpose Interactive Real-


Time Digital Image Processing
System

Avidan Akerib

Avidan J. Akerib, "Special-purpose Interactive Real-Time Digital Image


Processing System," Proc. SPIE 0579, Intelligent Robots and Computer
Vision IV, (11 December 1985); doi: 10.1117/12.950841

Event: 1985 Cambridge Symposium, 1985, Cambridge, United States

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use


Special-purpose
Special- purpose interactive
interactive real-time
real -time digital
digital image
image processing
processing system

Avidan
Avidan J.
J. Akerib
Akerib

Tadiran Electronic
Tadiran Electronic Industries
Industries Ltd.
Ltd. System
System Dept.
Dept.
26
26 Ha'Shoftim St. P.O.B.
Ha'Shoftim St. P.O.B. 267
267 Holon
Holon 58102
58102 Israel.
Israel. Telex
Telex 03
03-35413
-35413

Abstract
Abstract

This paper
This paper presents
presents aa new
new image
image processing
processing system
system intends
intends to
to enhance
enhance video
video images
images transmitted
transmitted
from remotely
from piloted vehicles
remotely piloted (R.P.Vs), or
vehicles (R.P.Vs), or any
any other
other airphotoes
airphotoes in
in real
real time.
time.
The objective
The objective was
was to
to select
select known
known algorithms
algorithms of
of image
image processing
processing techniques
techniques and
and to
to develop
develop special
special
purpose architecture,
purpose based on
architecture, based on pipe
pipe line
line processor
processor in
in hardware
hardware and
and top
top down
down structure
structure design
design in
in software,
software,
in order
in to apply
order to apply these
these algorithms
algorithms inin real
real time.
time.
The system
The system operator
operator is
is not
not required
required to
to have
have knowledge
knowledge of
of image
image processing.
processing. Interactive operations
Interactive operations
were
were chosen
chosen since successful image
since successful image enhancement is based
enhancement is based on the
the ability of
of the
the observer
observer to
to detect
detect infor-
infor-
mation
mation of
of interest in the
interest in the final
final process image.
process image.

The system
The system architecture,
architecture, the
the image
image processing
processing moduls
moduls and
and the
the data
data base
base management
management are
are described
described
briefly.
briefly.

Introduction
Introduction

Within the last


Within the last several
several years,
years, wide range of
wide range of real
real time
time image
image processing
processing systems
systems have
have been
been developed.
developed.
The
The most image processing
most image processing systems
systems design
design are
are based
based on
on special
special purpose
purpose hardware
hardware architecture,
architecture, and
and groups
groups
of image
of image processing
processing software
software are
are developed
developed around
around it,
it, enabeling
enabeling the
the user
user to
to design
design his
his specific
specific algorithms.
algorithms.
Those systems cannot
Those systems cannot be
be useful
useful if
if the
the operator
operator needs
needs immediate
immediate feedback
feedback andand is
is not
not experienced
experienced in
in image
image
processing, such
processing, such as
as in
in battelfield
battelfield conditions.
conditions. Systems
Systems that
that can
can answer
answer those
those cases
cases should
should be
be designed
designed
around higher levels
around higher levels architecture
architecture ofof software
software and
and should
should be
be more adaptive and
more adaptive and enough
enough intelligent
intelligent to
to under-
under-
stand
stand the
the requirement
requirement inin order
order to
to output
output an
an image
image with
with maximul
maximul useful information content
useful information content in
in real
real time.
time.
Such system
Such system is
is described
described below.
below.
Approach
Approach

The
The approach
approach for the system
for the system design was
was broken down into
into 4 steps
steps : :

1.
1. Define the
Define the mathematical
mathematical algorithms.
algorithms.
2.
2. Design the
Design the system
system architecture.
architecture.
3.
3. Realize the
Realize the mathematical algorithms in
mathematical algorithms in the
the system
system architecture.
architecture.
4.
4. Develop "intelligent"
Develop "intelligent 11 software
software that
that understands
understands the
the operator
operator requirement
requirement and
and calls
calls for
for the
the
specific image
specific image processing
processing moduls.
moduls.
The mathematical
The mathematical models include theoretical
models include theoretical algorithms
algorithms of
of image
image processing
processing which
which most
most of
of them
them were
were
developed
developed at
at previous
previous works; such as
works; such as box filters,
filters, feature select,
select, histogram
histogram equalization,
equalization, etc.
etc.
The system architecture
The system provides two
architecture provides two types
types of
of real
real time
time operations,
operations, either
either point
point or
or neighborhood
neighborhood
operations. Point operation
operations. Point operation consists
consists on
on transformation
transformation of
of each
each pixel
pixel value
value into
into aa new
new one.
one. ItIt is
is based
on updating
on updating anan output
output look
look up
up table
table during
during each
each vertical
vertical blank
blank period.
period. The neighborhood
neighborhood operations
operations depend
depend
on the
on the pixel
pixel itself
itself and
and its
its vicinity
vicinity also.
also. In
In this
this technique
technique the
the system
system has
has the
the ability toto memorize
memorize
previous
previous frames and performs
frames and performs linear
linear operations
operations upon
upon them.
them.
Realization
Realization of the algorithms
of the in the
algorithms in the system
system architecture
architecture is
is performed
performed by
by generating
generating structures
structures
and
and sub
sub structures
structures that
that each
each of
of them
them includes
includes aa calling
calling sequence
sequence of software
software drivers routines
routines to
to the
the
system hardware.
system hardware.

The software
The software development
development can interface between
can interface between the
the operator
operator requirement
requirement that
that used
used aa control
control box,
box,
and
and the
the system
system architecture
architecture such
such that
that for
for each
each user
user command,
command, aa sequence
sequence of instructions are generated
of instructions
automatically
automatically toto run
run on
on the
the system
system architecture
architecture to
to provide
provide the best image
image output. If there are some
output. If some
user's
user's commands
commands that
that are
are not
not fit
fit to
to image
image processing
processing rules (for example
rules (for - requirement
example - requirement for adaptive
contrast enhancement
contrast enhancement when previous request
when previous request was
was pseudo
pseudo color),
color), the
the software
software deletes
deletes all
all the
the previous
previous
commands that
commands that are
are not
not fit
fit to
to the
the last
last requirement.
requirement.

508 //SPIE
SPIEVol
Vol.579
579Intelligent
IntelligentRobots
Robotsand
andComputer
ComputerVision
Vision (1985)
(1985)

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
technique
image enhancement technique
The image
We shall here some of the image
shall describe here techniques used
image processing techniques used in the system.
in the system.

1. Filter
1. Filter

kinds of filters are available in


Two kinds the system
in the TIME
system -- TIME and SPATIAL
and SPATIAL

In Time Filter
In Time there is
Filter there an averaging
is an among successive
averaging among frames according
successive frames the following
to the
according to following
formula
formula

(1) Y(n)
(1) Y(n) = X(n) ++ (1(1 -- K)K) YY (n(n-1)
= KK X(n) -1)
original image
(n) == original
When XX (n) image
(n-1)
YY (n -1) = response to
Filter response
= Filter to previous frame.
previous frame.
00 << KK << 11
select the
The user can select parameter KK using
the parameter the output
ball and change the
track ball
using a track interactively
output interactively
he gets
until he
until desire.
his desire.
gets his
In Spatial Filter,
In Spatial are The
there are
Filter, there Pass, Band
The Low Pass, Pass and
Band Pass Pass Filters.
High Pass
and High Filters.

In LPF and
In LPF HPF,software generates
and HPF>software automaticaly aa mask
generates automaticaly to convolve
mask to the image.
convolve the image. The and
size and
The size
the constants
the that mask were
constants of that selected to
were selected be the
to be best to
the best fit the
to fit image. (Example
the image. of LPF
(Example of LPF
presented at
presented at Fig. 1)
Fig. 1)
In BPF, the
In BPF, operator can
the operator select the
can select spectral band
the spectral by using
band by track ball.
using aa track ball. Software then
Software then
generates according to
box filter according
generates aa box the ball
to the position and
ball position the output
and the output image obtained is
image obtained is aa
result of the two convolution
between two
the difference between of different
masks of
convolution masks sizes and
different sizes constants.
and constants.
All
All the selected to
masks were selected
the masks operate only
to operate columns, while
only on columns, the size
while the selects automatically
size selects automatically
software in
by software in aa range between 11 to
range between pixels. For
IK pixels.
to 16 For example band pass
example : a band : filter can
pass filter be
can be
aa result of with 1/4
masking with
of masking (1,2,1) T -- 1/8
1/4 (1,2,1) 1/8 (1,1,1,2,1,1,1) 128. Here,
(1,1,1,2,1,1,1) TT ++ 128. Here, there is aa
there is
sizes 33 and
difference between masks with sizes and 7.7. sizes were
The sizes by the
selected by
were selected ball
track ball
the track
selected automaticaly by
position and constants were selected
position the software.
by the of BPF
Example of
software. Example presented
BPF presented
B.
at Fig. 22 B.
2.
2. Feature select
Feature select
This operation classifies the
This image from 22 to
original image
the original each of
and represents each
groups and
to 8 groups them
of them
with aa different color. It
different color. achieved by
is achieved
It is computing the
by computing local minimum
the local the histogram
of the
minimum of histogram
"vallies".
"vallies ". Groups are selected by
then selected
are then integrating between
by integrating close vallies.
two close
between two vallies. If group
If aa group
includes than 12.5%
includes more than of the
12.5% of image histogram
the image histogram it is defined
it is as aa feature.
defined as feature. If not, it
If not, is
it is
added to the
added to group until
next group
the next sum has
the sum
until the the 12.5%.
passed the
has passed 12.5%.
Fig.
Fig. 3 represents the
3 A represents histogram.
original histogram.
the original
Fig. 3 B
Fig. B is selected group
the selected
is the Fig. 33 CC the
and Fig.
group and output gray
the output level.
gray level.
Fig. presents the
Fig. 2 C presents result of Fig.
select result
feature select
the feature 2 A.
Fig. 2 A.

A
A BB
Original
Original LPF of
of AA
Figure 11

SPIE Vol 579 Intelligent Robots


SPIE Vol. Computer Vision
Robots and Computer (1985)// 509
Vision(19851

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
AA -- Original
Original

/>;v*;^, <• .•-;,

BB -- BPF of
BPF of AA C -- Feature
C Feature Select
Select of
of AA
Figure
Figure 22

IIIINIIII IIIIIfiIIIIIIIIIIIIIIIIIIIIIIIIll II

*) ORIGINAL
AI ORIGINAL NISTOGRAN

GROUP GROUP GROUP GROUP,. ^


a 3

/
_
" \: \
v'
V
f |7X
V
"
''V- iox 1 sx
-*"
s~
^^_ /
9X ""1
I
1
j

34X
\
\

01 GROUP SELECTED IEACN APEA IS GREATER TNAN 21,

CI NEW NISTOGRANIA
Cl HISTOCPAMI4 FEATURES WERE
WERE SELECTED,
SELECTEDI

Figure 33
Figure

510
510 / /SPIE
SPIEVol
Vol.579
579Intelligent
IntelligentRobots
Robotsand
andComputer
ComputerVision
Vision(1985)
(1985)

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
3.
3. Edge enhancement
Edge
Here, there is
Here, there data base
is a data of 11
base of convolution's mask
11 convolution's
11 xx 11 for different
mask for textures. They
different textures. were
They were
chosen as such,
chosen that the
such, that center mask has
the center property of
has aa property point detector
of point and as
detector and far as
as far increasing
as increasing
the distance from
the distance the center,
from the the difference
center, the between the
difference between positives and
the positives the negatives
and the of the
negatives of the
either.
increases either.
mask increases
For
For example If the
example : If : left of
down left
the down the 11
of the mask is
11 mask
11 xx 11 located by
is located at point
definition at
by definition 0
point XX == 00 YY== 0
the masks at
the locations
at locations
(6,6) (11,11) (3,2)
(3,2) will be
will be
-1-1 1
-1 -1 1 -355 0-1-1
0 -1 -1
-1 8-1
-1 8 -1 -3
-3050 5 11 00-1
-1
-1-1-1
-1 -1 -1 -3-3-3
-3 -3 -3 1 11 00
1 respectively
respectively

Any mask
Any mask from the selected by
be selected
the field can be positioning the
by positioning track ball.
the track In this
ball. In principal,
this principal,
the gradient position is related to
position is ball position
the ball
to the gradient magnitude
the gradient
and the
position and magnitude is related
is related
the distance in
to the the track
in the motion.
ball motion.
track ball
There are two
two kinds be selected
can be
kinds of operations which can edge enhancement
by edge
selected by and
GRADIENT and
enhancement : GRADIENT :

SHARP operations.
operations.

At gradient operation, software adds


the software
operation, the the value
adds the 128 to
value 128 to the result of
the result of the convolution
the convolution
described above, while at
described above, Sharp operation
at Sharp the original
operation the original image is added
image is added to result.
the result.
to the
Example of Grad and Sharp
Example operations at
Sharp operations at location illustrated at
is illustrated
(0,0) is
location (0,0) at Fig. and 44 CC
Fig. 44 BB and
respectively.
respectively.

+
t
tS t.
A - Original B of A
B -- Grad operation of

of AA
Pseudo color of
0 -- Pseudo of
C -- Sharp operation of A
A
Figure
Figure 4
4

SPIE Vol 579 Intelligent


SP1E Vol. intelligent Robots Computer Vision
Robots and Computer (1985)// 511
Vision{1985}

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
There
There are
are many other available
many other available image
image operations
operations in
in the
the system,
system, such
such as
as the
the adaptive
adaptive contrast
contrast
enhancement, zooming, pseudo
enhancement, zooming, pseudo color,
color, thresholding
thresholding conturing,
conturing, window operations, etc.
window operations, etc.
Fig. 4 DD presents
Fig. 4 presents pseudo
pseudo color
color of
of Fig.
Fig. 44 A.
A. Fig.
Fig. 55 AA presents
presents contrast enhancement inside
contrast enhancement inside
a window.
window. TheThe enhanced window was
enhanced window was split
split down
down for
for comparison.
comparison. Fig. 5 BB presents
Fig. 5 presents zoom
zoom
operation of
operation of 55 A.
A.
There is aa possibility
There is possibility toto generate
generate numbers
numbers of
of image
image operations and
and the
the software
software will
will decide
decide
on the best
on the best sequence
sequence of
of command
command to
to operate
operate the
the image
image according to
to the
the rules
rules of
of the
the system
system
programmi ng.
programming.

A BB
Contrast enhancement Zoom operation
Zoom operation of
of A
A
inside aa window
inside window
Figure 55

Hardware
Hardware

As
As stated above, the system
above, the system design
design is
is intended
intended to
to perform
perform real
real time
time image processing, The
image processing.
algorithms
algorithms discussed
discussed above
above are
are divided
divided into
into two
two generic types of
generic types of processing.
processing.
a)
a) Point
Point processing
processing -- The
The simplest type
type -- consisted
consisted of aa transformation of each
each original
original
image
image pixel
pixel value into
into aa new value
value for
for the
the output image,
image, such as
as adaptive contrast
enhancement,thresholding, etc.
enhancement,thresholding, etc.

b)
b) Neighborhood
Neighborhood processing
processing -- performs
performs aa transformation
transformation on
on each
each pixel in aa way that depends
pixel in depends
not only
not only on
on the
the gray
gray levels
levels of
of the
the pixel
pixel being
being processed,
processed, but
but also
also on
on its
its vicinity
vicinity and
and
includes
includes techniques
techniques such
such as
as edge
edge enhancement
enhancement and
and spatial
spatial filtering.
filtering.

All
All the
the above,
above, are implemented
implemented by
by using
using the
the system
system described
described at fig. 6.
at fig. 6.
The
The digitizer includes two
digitizer includes two sections
sections : The A/D section
section samples
: the image,
samples the image, and the
the D/A section
section
displays it.
displays it. The
The digital image feeds
digital image the D/A through
feeds the through aa look
look up
up table (LUT)
(LUT) in
in order
order to
to perform
perform
a discrete transformation
transformation on
on the
the digital
digital data
data before
before it
it is
is displayed.
displayed.
At point operations,
operations, the
the digital
digital input
input obtained
obtained from
from the
the AA/D,
/D, feeds the LUT
feeds the LUT and
and the
the D/A
D/A
via the
via the pipe
pipe line
line processor.
processor. The
The mini
mini computer
computer which
which controls the system
controls the system can
can update
update the
the LUT
LUT related
related
to
to the
the required
required operation.
operation. The LUT updating
updating occurs
occurs only at vertical
vertical blank period
period to
to prepare
prepare the
the next
next
field for aa new
new operation.
operation.
At neighborhood operations
At neighborhood the system
operations the system architecture
architecture should
should have
have the
the ability to memorize
ability to memorize previous
previous
frames
frames and^performs linear operations
and performs linear operations among
among neighborhood pixels. For
neighborhood pixels. For those
those cases
cases there
there was
was aa need
need to
to
grabb
grabb the
the input image at frame storage,
input image storage, performs
performs neighborhood
neighborhood operations
operations on it by
on it by using pipe line
using pipe line
processor, then
processor, then to
to display
display the
the result
result on
on the
the monitor.
monitor. 44 frame
frame storages
storages are
are available
available in
in that
that system:
system:

512 //SPIE
SPIEVol.
Vol. 579
579Intelligent
IntelligentRobots
Robots and
andComputer
Computer Vision
Vision (1985)
(1985)

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Input Image
Input image storage,
storage, output
output image,
image, and
and other
other two
two are
are configured
configured together
together into
into 16
16 bit
bit deep
deep to
to perform
perform
the
the image
image processing
processing in
in order
order to
to avoid
avoid overflow
overflow at
at processing. Histogram card
processing. Histogram card was added in
was added in order
order to
to
obtain
obtain the image dynamic
the image dynamic range
range from
from the
the previous
previous frame.
frame.

TV CAMERA
^ TV MONITOR

The real time


The real time hardware
hardware architecture
architecture
Figure
Figure 6
6

Software
Software

The software
The software system
system was written down
was written down in
in aa "top
"top down"
down" structure
structure design
design modular manner.
modular manner.

The main
The main structure
structure modules are shown
modules are shown at
at Fig.
Fig. 7.
7. We shall
We shall describe
describe here
here their
their performance.
performance.

INIT

4
MAIN
LOOP INTER

i
11 <1
IMAGE IMAGE
PROCESSING MANAGEMENT

i i 1 , i 4 w \1 W
FILTER
FILTER c4*hE STRETCH
STRETCH CONVOLV
CONVOLV COLOR
COLOR WINDOW
WINDOM
DISC
DISC
CONTROL
CONTROL
ZOOM
ZOOM

»
TIAC
.M.E
1M
•BIT
IRE

kV
E11 SLICE
.1(111 III
MIt EOU
•HIST
1100
EOU
•LINE.6TRE
L IRE .EIE
RHO
:!5?°
SMtV
LRF -
COLM ISTOR89w
»ECT
DÌRECT

The software
The software structure
structure
Figure
Figure 77

SPIE Vol.
SPIE Vol 579 Intelligent Robots
Robots and Computer
Computer Vision
Vision(1985)
(1985)// 513

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
1.
1. Init :
Inlt : It Includes
includes the first submoduls
the first be executed
to be
submoduls to after the
executed after operation. It
boot operation.
the boot initia-
It initia-
lizes all variables
lizes all addresses, and
and addresses,
variables and self test
executes aa self
and executes the hardware.
to the
test to hardware.

2.
2. Inter : : It interrupts service
includes 3 interrupts
It Includes routines :
service routines :

Control box interrupts


Control its usage is
interrupts -- its read the
to read
is to user requirements.
the user requirements.

Vertical blank -- its


Vertical blank usage is
its usage updating the
for updating
is for LUTs.
the LUTs.
image histogram
Histogram execution -- The image found and
was found
histogram was and is to be
ready to
is ready read.
be read.
3. loop:
Main loop:
3. Main It includes
It the following
includes the operations :
following operations :

Reading the image processing operation from


the image the control
from the box.
control box.

- Checking the type


Checking the of the
type of selected operation
the selected and executes
operation and it by
executes it one of
by one the
of the
following modules : :

a)
a) Image processing
b)
b) Image management
Image

"menu" on
of "menu"
Management of screen.
the screen.
on the
to monitor aa processed
Outputs to image.
processed image.
4.
4. Image processing:
Image pipe line
generates aa pipe
processing: This module generates includes the
link which includes
line link processing
image processing
the image
operation to
operation be executed
to be the following
by the
executed by steps :
following steps :

a)
a) It receives the
It receives image operation
the image requirement and
operation requirement fixes its
and fixes to the
entrance to
its entrance the
queue of the pipe line
queue link according to
line link the programmed
to the rules.
programmed rules.
b)
b) Calls its operation to
each module for its
Calls each its order
to its
to execute according to the
in the
order in
queue.
queue.

c)
c) Transfers to parameters.
its parameters.
to each module its
5.
5. Image management: This
Image management: module receives
This module data management from
image data
receives the image the "main
from the loop"
"main loop" and transfers
and transfers
it to
it following 33 modules
the following
to one of the control or
Disc control
modules : Window, Disc : zoom.
or zoom.

IMIiMfM1111 D2MECTO1Mt

EMBIMMONEMININIgniiiiiMMEMMMON
e¢ "
e'
27_

Figure 8
8

Vol.579
SPIE Vol
514 //SPIE IntelligentRobots
579Intelligent Computer Vision
andComputer
Robotsand (1985)
Vision (1985)

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
In case
In case of
of recording image on
recording image on aa disc (suboption of
disc (suboption the disc
of the disc control),
control), the
the main loop receives
main loop receives
time
time and
and date
date from
from the
the operating
operating system
system to
to provide
provide images
images headers
headers for
for directory
directory file.
file.

Images
Images directory
directory display
display is
is an
an another
another suboption
suboption of
of disc
disc control
control and
and is
is presented
presented at
at Fig.
Fig. 8.
8.

Operation
Operation

The
The control
control box
box described
described at Fig. 99 enables
at Fig. enables the
the user
user to
to select
select his
his requirements.
requirements.

The"Menu
The "Menu on"key
on "keyselects
selectsa amenu
menuofofrequirements
requirements at
at the
the right
right bottom
bottom of
of the image.
the image.

The INC and


The INC and DEC
DEC keys,
keys, position an arrow
position an arrow cursor
cursor pointed
pointed to
to one
one of
of the
the menu
menu options.
options.

The
The Sel
Sel key
key selects
selects aa suboption
suboption defined by the
defined by the cursor,
cursor, and
and aa new
new menu is presented.
menu is presented. If
If the
the Select
Select
key
key is
is pressed
pressed on
on the
the last
last suboption,
suboption, the
the required
required process
process is
is generated
generated and
and the
the main
main menu is presented,
menu is presented,
showing
showing the
the processing status.
processing status.

MENU
MENU
ON
MENU
MENU
OFF
OFF

SEL

/NCR
I NCR PREY
PREV DECK
OECR

FRZ ZOOM

TRACK BALL
TRACK

MENU SELECT

LOCAL AKgA
HISTOGRAM SQOALIZATIOH
rim*

The control
The control box
box
Figure 9
Figure 9

Conclusions
Conclusions

AA new
new real
real time image processing
time image processing system
system was presented. Combination
was presented. Combination between
between high technology
high technology
hardware architecture,
hardware architecture, intelligent
intelligent software
software and
and useful
useful image
image processing
processing algorithms
algorithms can
can help
help the
the
unspecialist operators
unspecialist operators in
in real
real time
time image
image understanding..
understanding..

Acknowledgement
Acknowledgement

I would
I like to
would like to thank
thank "Tadiran
"Tadiran Electronic
Electronic Industries
Industries Ltd.,
Ltd., System
System Department
Department", Holon, Israel,
", Holon, Israel,
for
for their
their great
great assistance
assistance byby helping
helping to
to bring
bring this
this theoretical
theoretical work
work into
into actual
actual practice. This
practice. This
system was
system was produced in the
produced in the laboratories,
laboratories, and
and by
by the
the equipments
equipments and
and unlimited
unlimited budget
budgetofof"Tadiran
"Tadiran".
".
The
The experiments lasted almost
experiments lasted almost year
year under
under the
the instruction
instruction of
of Professor
Professor N.N. Rosenberg
Rosenberg ofof Tel
Tel-Aviv University.
-Aviv University.

SPIE Vol.
SPIE Vol 579 Intelligent Robots Computer Vision
Robots and Computer Vision(1985)
(1985)// 515

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
References
References

1.
1. Corbett
Corbett Francts
Francis && Rashtd beg, Operations
Rashid beg, Operations && Architectures
Architectures Developed
Developed On
On The
The IPIP-512 Image
-512 Image
Processing family. The
Processing family. The International
International Society
Society of
of Optical
Optical Engineering,
Engineering, Vol. 504, Aug.
Vol. 504, Aug. 84.
84.
2.
2. Green
Green William B., Digital
William B., Image processing,
Digital Image processing, Van Nostrand Reinhold
Van Nostrand Company Inc.
Reinhold Company Inc. 1983.
1983.
3.
3. Hanaki
Hanaki S.,
S., An Interactive Image
An Interactive Image Processing
Processing && Analysis
Analysis System,
System, from
from real
real time
time/parallel
/parallel
&& image
image analysis,
analysis, Pelenum Press, New
Pelenum Press, New York
York 1981 pp. 219
1981 pp. 219 -- 225.
225.
4.
4. Haralick,
Haralick, Robert
Robert M.,
M., Digital
Digital Step Edge from
Step Edge from Zero
Zero Crossing
Crossing of Second Directional
of Second Directional Derivatives,
Derivatives,
IEEE Transactions
IEEE Transactions on
on Pattern
Pattern Analysis
Analysis && Machine Intelligence, Vol.
Machine Intelligence, PAMI -- 66 No.1
Vol. PANI No.l Jan.
Jan. 84.
84.
5.
5. Patrenaili M. Narendram
Patrenalli M. & Robert
Narendram & Robert C.
C. Fitch,
Fitch, Real
Real Time
Time Adaptive
Adaptive Contrast Enhancement, IEEE
Contrast Enhancement, IEEE
Transactions on Pattern
Transactions on Pattern Analysis
Analysis && Machine Intelligence, Vol.
Machine Intelligence, Vol. PAM
PAM 11 -- 33 No.
No. 6,
6, Nov.
Nov. 1981
1981
pp. 665
pp. 665 -- 670.
670.
6.
6. Pratt W.
Pratt W. K.,
K., Digital
Digital Image
Image Processing,
Processing, New
New York,
York, Wiley
Wiley -- 1978
1978
7.
7. Proceeding of
Proceeding of SPIE,
SPIE, The
The International
International Society
Society for
for Optical
Optical Engineering
Engineering Vol.
Vol. 301,
301, Design
Design of
of
Digital Image Processing
Digital Image Processing Systems
Systems Aug.
Aug. 1981.
1981.
8.
8. Shorki Z.
Shorki Z. Selim
Selim && M.A.
M.A. Ismail,
Ismail, KK means - type
means - type algorithms,
algorithms, AA Generalized
Generalized convergence
convergence Theorem
Theorem
& Characterization
& Characterization of
of Local
Local Optimality
Optimality ,, IEEE
IEEE Transactions
Transactions on Pattern Analysis & Machine
Intelligence,
Intelligence, Vol. PAM 11 -- 33 No.
Vol. PAM No. 6,
6, Jan.
Jan. 1984
1984 pp.
pp. 81
81 -- 86.
86.

516 //SPIE
SPIEVol
Vol.579
579Intelligent
IntelligentRobots
Robotsand
andComputer
Computer Vision
Vision (19851
(1985)

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 12 Sep 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

You might also like