Design and Research on Touch-sensitive Image Measuring Instrument with High Precision

beaverswimmingAI and Robotics

Nov 14, 2013 (3 years and 7 months ago)

53 views



1

Design and Research on Touch
-
sensitive Image Measuring Instrument
with High Precision

Yinda

Sun

1
,

a
,

Wen
q
in

Xu

2
, b
,

Baiqing Zhou
3,c

and F
eng

Z
hu
4
,
d

1
Zhejiang Industry Polytechnic College, Shaoxing,
312000
,

PR

China

2
Zhejiang Industry

Poly
technic College,
Shaoxing,
312000
,
PR

China

3
Huzhou Vocational and Technology College,
312000
,
PR

China

4
Zhejiang Industry Poly
technic


College, Shaoxing,
312000
,
PR

China

a
ssss75828@126.com
,
b
l
unaseen@126.com
,

c
zhoubaiqing@hzvtc.net
,

d
Z.F
-
1972
@1
63
.com

Keywords:

touch
-
sensitive monitor, system calibration, image detecting
algorithm
s, touch
-
sensitive
image instrument.

Abstract.

A digital image detecting system that employs a touch
-
sensitive moni
tor as
operating

center is stated here.
W
ith touch
-
screen and
easy
-
to
-
use measuring software, its efficiency has been
greatly improved compared to the
traditional

manual image instrument.
I
ts basic structure and
measuring principle are presented first,
and

then

the software

s interface and the calibration method,
and finally the fundamental principle of the two key algorithms for image detecting: sampling and
marking.

Introduction

With

the rapid progress of computer, optical technology and image
technology
,

the optical image
measuring system has been widely used to measure parts and to detect the size of products and
semi
-
finished products in industries of machine, mold, and electron.
I
t has a precise result and an
intelligent procedure as well. Instruments
from
Mitutoyo
,
TESA
, and Yixin Suzhou are typical ones
which
are classified

into two types in general: the automatic and the manual.
T
he former is powerful
in which the detecting personnel can not only drive the motor to work but also program it to detect
automatically.
I
t has a higher
manufacturing

cost and price because motor and motion control card are
employed, which are used in measurement laboratory. While the latter has a lower efficiency and
price because the working platform is needed to be moved b
y manpower and the measuring functions
are needed to be chosen by operating the mouse in the computer

s
interface,
which is suitable for
detecting
small

quantities of workpieces.

B
ased on the digital image processing
technology
, t
he
touch
-
sensitive image
measuring instrument
uses touch
-
sensitive monitor as the interactive center.
T
he operator measures the workpiece by
choosing buttons on the screen after moving the working platform, and thus reducing the operation of
mouse.
M
eanwhile, it has a higher effic
iency
because

a touch
-
pen is used to do marking measurement
directly along the image border on the screen after choosing corresponding measurement button.
B
esides, it has a lower manufacturing cost because motor and motion control card are omitted.
T
he
mea
suring accuracy is up to 5

m because of the automatic tracing
-
edge technologies of detecting
sub
-
pixel, segmenting region, and of compensating precision.
F
urthermore, the features of being easy
to operate and having short operator

s t
raining
-
time make it fit for size detecting in production line.

Hardware’s

Structure and Basic Principle

Figure 1 is
a

structural diagram of the touch
-
sensitive image measuring instrument. Figure 2 is
a

real
graph.




2




1. Touch
-
sensitive monitor 2. Movin
g slider in Direction Z 3 CCD 4. Microscope

s lens 5 Upper lamp source 6.
Column in Direction Z 7. Industrial computer 8. X
-
Y worktable 9. Lower lamp source 10. Granite platform

Figure 1. Structural Diagram of the Touch
-
sensitive Image Measuring Instr
ument

As shown in Fig. 1, put the workpiece to be
detected

on the platform and turn on the upper or the
lower lamp, the workpiece is imaged on the target surface of the CCD through the microscope

s lens.
T
he optical signal is then converted into electrical

signal by CCD, and finally is
transferred
to the
image acquisition card in computer, the digital signal then displays on the touch
-
sensitive
monitor.

T
he acquisition system has two lighting modes: upper light and backlight.
T
he
former

is composed of
high
-
bright LEDs that are assembled annularly at the lamp

s shell according to certain angle and
distribution density that are decided by the lens working distance and the illumination needed to
gather image by CCD; the latter is made up of one high
-
brightness

high
-
power LED that locates in the
lens


focus.
T
he beam emerges in parallel from the lens, shining onto the workpiece through the
diaphanous

glass plate. The measurement error caused by
image

is eliminated
dramatically

with the
uniform backlight.

T
he sy
stem uses a zoom lens of 0.7X
-
4.5X from Guilin
Optics

with

an amplified factor of up to
4.5x,

an working distance of 90mm, an analog CCD camera from MINTRON with a target surface
size of 1/2 inch, an image acquisition card (from Tianmin) of Model SDK3000 t
hat has a frame rate of
40 frames/s and a
maximum

analytical pixel of 768

576, and works in PAL mode.

W
hen the system has an amplified factor of 4.5X, it has a minimum visual field of 2mm

1.5mm. The
maximum
resol
ution

is computed by:

P =
n
X
dim
=
576
1000
5
.
1


2.6üm

(Where, dimX is the size in Direction X, n is the analytical pixel in Direction X, P is the resolution)



Figure 2. Real graph of the Touch
-
sensitive
Image Instrument

T
he measuring accuracy in single image is usually decided by the resolution image, but
affected
greatly

by the grating ruler when measuring big size workpieces because the microscope has a narrow
visual

field.
H
ere, a grating ruler
with

th
e resolution of 1

m from Guiyang Xintian is employed to


3

improve its accuracy through a linear compensation by a glass line scale at the maximum optical
amplified rate.
T
he scale

s reading is
just

the compensation standard.

W
hen the c
omputer obtains a steady and clear image, the user will measure it by operating the
software

s buttons on the touch
-
sensitive monitor.
T
he measuring software will have
corresponding

points group by technologies of tracing edge automatically, segmenting reg
ion,
analyzing

sub
-
pixel
and so on.
G
eometry sizes and parameters of circle, line segment, rectangle and gear are thus obtained
by substituting the corresponding fitting function according to the measurement needs.

Design the Software Interface

T
he touch
-
sensitive
monitor

is a communication and interactive medium between user and computer.
B
ased on its response characteristic, buttons and interfaces are designed to be humanized and easy to
use.
T
he basic interface is shown in Fig.3.

U
sers observe the wor
kpiece image from the image window
that

located in the left region of the
interface.
T
o enable the operator to recognize the workpiece

s characteristics, the window is designed
with a maximum
analytical

pixel of 768

576. A relevant
algorithm

is developed
s
pecially

to play the
touch screen


role better and to improve the measurement efficiency. When choosing the geometric
figure to be
measured

in the interface, the operator draws a figure on the border of the image
with

touch pen.
T
hen it is convenient for t
he software to obtain the measurement result.
That

is, what you
see is what you draw.
T
he prompt message and measurement result is shown in the lower region of the
window, where users can operate according to the prompts and can see the results when finish
ing
measuring
.
T
he top
-
right corner is an X
-
Y
coordinate

that exhibits the
corresponding

values got from
the grating ruler.
U
sers can do a contrastive measurement by moving
the

worktable without the
accurate result. The lower region of the coordinate

displ
ays f
unction
buttons such as circle, line
segment, distance, angle, rectangle, gear
and

crew.
Big
buttons are designed for an easy operating on
hand screen or by touch pen.




Figure 3. Operation Interface of the Software in the Touch
-
sensitive Image In
strument

Image

s C
alibration

of the Detecting System

C
alibration

is an important job for machine vision in size detecting.
T
he detecting error will be caused
by CCD camera itself, or/and by coordinate conversion between image and space, or/and by external
environmental factors.
A
n optical lens of 1X
-
4.5X (from Guilin
Optics
) that has small distortion is
adopted

to meet the requirements of liner
calibration
.

As shown in Fig.4, the image is
calibrated

by a standard circle from the precise glass standard
t
emplate that has been checked by a high
-
accurate measuring instrument.
A

contour edge is obtained
by extracting
contour [
1
,

2
,
3
]
, and then an accurate coordinate of boundary points and a circle center
are obtained by making

a
gradient analysis search for
t
he

eight neighborhood sub
-
pixels [4
,5
]
.
T
he


4

instrument is calibrated accurately from the corresponding calibration coefficient by comparing
the
true value with
the pixel equivalence because the true diameter is known.




Figure 4 Image

s Calibration and

Glass Standard Detecting Template

Usually, whether the calibration circle is suitable or not has some influence on the precise of the
image measuring system. In the template, there are circles of diameters of 0.25mm, 0.5mm, 0.75mm,
1mm
, 2.5mm
, and 5mm. Ju
st choose the
calibration circle
that
is
one
quarter

of the image window.

Basic Principle of the Image Detecting
Algorithm

Two methods are used to measure geometric figure: sampling and marking.
T
he detecting procedure
is shown is
Fig.5
. Fig.6 is the
instr
umentation plan
.














Figure 5 Flowchart of the Detecting
Algorithm

T
he sampling method is widely used, in which with the interval of 40ms, the computer draws cross
center lines in the center of the touch screen
con
stantly.
M
ove the worktable, ali
gn the cross center
line to the points to be measured, and press the confirmed button in the software

s interface.
T
hen the
computer draws a detection line of rectangle array based on the center of the starting point that will be
made an accurate eight nei
ghborhood search. Seek the gradient of the gray value for the adjacent
points in all directions, and finally compute its extreme value.
P
osition of the
extreme

value is just the
precise boundary
position
.
When

the
coordinate

of the optimal boundary is obta
ined, the world
coordinate is obtained through coordinate transformation. Move the other points, press the
conformed button to repeat the procedure above to record the coordinate.
S
ubstitute the coordinate
points into the image
-
computing function and get
the geometry size.

T
he drawing method is mainly used to measure a complete geometry image in sight.
C
hoose
function button on the touch screen, and draw a corresponding shape along the border of the image by
touch pen.
T
he computer will draw a rectangle d
etection array for each point based on the drawn
border and search for an optical boundary point.
A
fter the world coordinate is transformed and all the
points is detected, substitute the points into the computing function for fitting (such as least square
To transform world
coordinate

for the boundary point

T
o search from the center of the
cross line

T
o draw a detection line

of
rectangle

array

based on
the center of the starting point

T
o obtain an optimal border through detection
analy
sis



5

method, linear regression method [6
,7
]
), by which, the relevant geometry size and parameter of line
segment, rectangle, distance, and centroid are thus obtained.











Figure 5 Instrumentation Plans of Sampling and Drawing

Conclusion

T
he touch
-
sensit
ive image measuring instrument based on digital image processing integrates touch
screen technology, optical image technology, digital image technology, mechanical manufacturing
technology

and software
technology
. W
ith

the measuring accuracy of up to 1um a
t the maximum
optical rate, it is an innovation base on the traditional image measurement instrument and
projector
.

W
hen the instrument is exhibited in the Beijing International Machine Exhibition, dealers and users at
home and abroad show their ken intere
st. Judging from sales and users


tests, the instrument is of high
precision, good
reliability
, and high efficiency.
F
urthermore, the self
-
developed software can
customize

SPC statistical program for users, which results in a lower
manufacturing

cost and a
n
industrial extension.
In
days to come, the main development approach lies to a better
customized

service and a
convenient

and fast operation.

Acknowledgment

This work was finacially supported by Science and Technology Planning Projects, Zhejiang
Province
(2009F70043)
.

Reference

[1]
Pal N and Pal S. A review on image segmentation

techniques. Pattern Recognition, 1993, 26(9):
1277

1294.


[2]

Malpica N, Ortuño J E, and Santos A. A multichannel

watershed
-
based algorithm for supervised
texture

segmen
tation. Pattern Recognition Letters, 2003, 24(9
-
10):1555

1564.

[3]
Canny J. A computational approach to edge detection. IEEE

Trans. on Pattern Analysis and
Machine, 1986, 8(6):679

698.


[4] Yujin Zhang. Image segmentation. Beijing: Science Press
. 2001
.


[5] Yujin Zhang. Image Engineering ( Volume 1) Image Processing and Analysis .P205
-
206 . B
eijing:

T
singhua
U
niversity

P
ress
. 2003
.

[6] Yetai Fei. Error Theory and Data Processing. Beijing: China Machine Press. 2003.

[7] Tiebang Xie.
Interchangea
bility

and

Technical Measurement. Wuhan: Huazhong University of
Science and
Technology

Press.1999
.





6