Note: Descriptions are shown in the official language in which they were submitted.
13~12~ 1
Field of th_ Invention
The present invention relates to web texture sensing.
More particularly the present relates to texture sensors applying
image analysis to instantaneous image signals generated by a
video camera operated in synchronization with a strobe light.
Background to the Invention
Many different instruments have been designed for on-
line sensing or web characteristics. Such equipment particularly
related to paper includes on-line sensors for paper formation,
fibre orientation, wire mark, etc.
One o~ the earlier developments for detecting formation
is described in a paper "A Formation Tester Which Graphically
Records Paper Structure" by surkhard et al published in the Pulp
and Paper Magazine of Canada, 60, No. 6, T319-T334 (June 1960)
wherein it was emphasized that direction was an important factor
that had to be considered in an instrument for measuring paper
structure. This device utilizes an illuminator and a photo
electric cell to sense the amount of light transmitted through
the web and generates a transmitted light signal which is
analy~ed in a frequency spectrum analyzer. The light source and
photo electric cell move axially of a cylinder on which the paper
sample is mounted.
A more recent system is described in an article
entitled 'Development o~ an On-line Formation ~ester to Determine
Optimal Use of Retention Aids' by Landmark et al published in the
Paper Trade Journal, September 1984, pages 84 to 86. This system
is based on the use of tester incorporating a laser and that was
developed by the Norwegian Pulp and Paper Institute to measure
the variations in basis weight. In particular for the instrument
described by Landmark et al a specific laser was selected and
illuminated a spot about one millimeter in diameter by directing
light through the web, the light transmitted through the web was
then projected onto a photo cell.
In U.S. Patent 4,648,712 issued March 10, 1987 to
Bernholt a light source illuminates one side o~ the web and the
i~k
13~0~8
radiation passing through the web is detected by a pair of
detectors having fields of view along narrow strips which are
perpendicular to each other. The detector outputs are processed
to provide fibre orientation ratio, formation and basis weight
measurements.
U.S. Patent 4,760,271 issued July 26, 1988 to sernholt
describes a system, which like the above described sernholt
system, illuminates paper from one side and provides optical
means on the opposite side for detecting the light transmitted
through the paper. A second illuminator is provided on the same
side of the paper as the detector and directs light onto the
paper at an oblique angle to a line normal to the plane of the
paper web and the reflected light emanatin~ from the second
source is also detected.
U.S. Patent, 4,64~,174 issued February 17, 1987 to
Ouellette et al also describes directing a beam through the paper
and utilizes a photo detector receiving the light transmitted
through the paper in combination with a tuneable filter and a
demodulator to produce a DC output reflecting size and
distribution of fibers or flocs.
It is also known to measure the surface properties of
paper by optical measurements. A recent technique for such
optical measurement is described in a paper entitled 'Optical
Measurement Throws New Light on Paper Surface' by Hansuebasi and
Morantz published in Paper Technology in Industry, August 1987.
This technique employs a scanning goniometer which varies the
angle of incidence of impingement of a polarized light and
studies the angular reflectance and attempts to correlate
printability with surface non-uniformity as determined by the
3Q equipment.
Equipment has also been proposed for surface inspection
for defects wherein the entire surface web is scanned to detect
defects or faults in the paper and enhanced graphics are used to
visually display a defect such as a small tear or the like in the
paper. Such a device described in a paper entitled 'Solid State
Surface Inspection Equipment' by P.W. Loose published in June,
' ~310128
1987 Tappi Journal, pages 69 to 74. This ~system uses a line scan
camera having a charged coupled device (CCD) and an electronic
exposure control. The amount of light applied to the CCD of the
camera is regulated and the basic scanning rate of the camera is
synchronized with the web speed.
An on-line dirt counting systems is described in an
article by Kemeny et al. entitled 'On-line Automatic Pulp Dirt
Count Measurement' Tappi Conference proceedings, 1987, pages 21
24 inclusive. The particular dirt counter described in this
publication utilizes the CCD camera focused onto an illuminated
surface of the pulp so that the light saturates the scanned
segment of the pulp surface to eliminate any shadows which may
tend to appear due to the roughness of the pulp surface. This
device determines the dirt count by correlating the signal
strength with a sase Line following threshold which was adjusted
so that the system tracks around the normal product fluctuations
and detects primarily sharp edge defects that generate a signal
sufficiently high to trigger the system.
Brief Description of the Present Invention
It is the main object of the present invention to
provide a surface te~ture sensor to determine on-line the
printability of a paper, for example as it is being produced.
It is a further object of the present invention to
provide an optical system for on-line determination of formation
of a paper web on-line.
Another object of the present invention is to provide
an on-line monitor for the detection of the degree of calender
blackening of a moving web. Calender blackening appears as
translucent specks on the web caused by over densification of
the paper structure in the calender nip.
It is a further object of the present invention to
provide an optical on-line pulp sheet dirt monitor.
Broadly the present invention relates to an optical
system for determining optical characteristics of a travelling
web comprising strobe light means for instantaneously
`` ~31~2~
illuminating areas on an exposed surface of said web as said web
traverses said strobe light means, video camera means focussed to
view said areas when illuminated by said strobe light, means
synchronizing the operation of said strobe with the operation of
said camera to provide a video siqnal for each of said areas and
means to analyze said signals to determine characteristics of
said web.
Preferably said strobe light will direct a beam of
light through said web from a second surface of said web said
second surface being opposite said surface.
Most preferably said strobe light will project a beam
of light onto said exposed surface of said web and will further
include means for perpendicularly polarizing said light in a
direction parallel to the plane of travel of said exposed surface
thereby to illuminate each said area with light polarized
parallel to the plain of said paper, said beam of light being
directed to said paper at an angle within three (preferably less
than one) degrees of the srewster's angle for the material from
which said web is made and said camera has its focal axis aligned
to receive light reflected from said area at the specular angle
of said polarized light and polarized in substantially the same
plane as said polarized light projected onto said surface.
Preferably the longitudinal axis or direction of said
beam of light and the focal axis of the camera will be in a plane
perpendic~lar to the direction of travel of the web.
Preferably for determining printability the size of
each said area will be coordinated with the size of the gloss
mottle elements to be considered. For newsprint areas in the
order of about 3 to 10 square centimeters are effective.
3~ Said camera may be any suitable black and white or a
color video camera preferably using a charge coupled device (CCD3
as its pick-up.
To determine the printability of said web the signal
from said video camera generated longitudinally specular
polarized light will be digitized and a histogram formed of the
degrees of brightness and the histogram analyzed to determine the
:
` ~31~128
distribution for degrees of brightness, a pluralitv of such
distributions accumulated and a portion of the paper represented
by said areas then classified as to surface texture.
To determine formation and blackening, the strobe light
illuminates the web from one side of the web with the video
camera directed at the opposite side of the web. Preferably the
focal axis of the camera will be perpendicular to the surface of
the web.
To determine the dirt count on a pulp sheet preferably
the camera will be positioned with its focal axis perpendicular
to the surface of the web and the web will be illuminated by a
strobe light on the same side of the web as the camera so that
the camera generates instantaneous signals o~ areas on the
surface of said pulp sheet~ The instantaneous signals are
digitized and analyzed to determine number and preferably
location of pixels in the camera illuminated below a certain
threshold to provide a dirt count for the pulp sheet.
A Brief Description of the Drawings
; 20 Further features, objects and advantages will be
evident from the following detailed description of the preferred
embodiments of the present invention taken in conjunction with
the accompanying drawings in which
Figure 1 is a side elevation schematically illustrating
a surface texture sensing device particularly useful in
determining printability on-line.
Figure 2 is a plan view illustrating the illuminated
spot and the area viewed by the camera relative to the spot.
Figure 3 schematically illustrates a system for
; 30 traversing the sensor laterally of the direction of travel of
paper web.
Figure 4 is a schematic illustration of a suitable
system for operating the present invention.
Figure 5 illustrates a histogram of digiti2ed light
intensity data for signals of pixels in a single video frame
developed by the camera.
~L31~ i 2$
Figure 6 illustrates the correlation between detected
gloss variation as determined by reflected light intensity
distribution and a measure of conventional MTI print mottle of
solid black prints.
Figure 7 illustrates a device for measuring formation
and/or blackening and in dot/dash lines a device for detecting
dirt.
Figure 8 in a plot of Formation as determined by the
present invention against Toyoseiki total formation to illustrate
the correlation.
Preferred Embodiments of the Present Invention
Surface Texture or Print bllity Sensor
One of the most important forms of the present
invention is illustrated in Figures 1 to 6 inclusive and utilizes
a system of sensing polarized light reflecting off the surface of
the web to obtain a measure of large scale topography that
effects print density uniformity. In particular the system
schematically illustrated in Figure 1 includes a strobe light 10
that directs a beam of light schematically indicated at 11 and is
triggered via a strobe trigger system that synchronizes the
operation of the light 10 and the video camera 14 as will be
described in more detail herein below.
In the arrangement illustrated the beam of light 11
from strobe light 10 is polarized by a polarizing system 16 that
polarizes the light perpendicular to the direction of the beam
and in a direction parallel to the surface 18 of the paper web or
the like 20 being scanned as indicated by the dots in the circle
22 indicating the light is polarized in the direction into the
paper. This polarized light is directed to the surface 18 at an
angle A which is substantially equal to srewster's angle for the
substrate 20 being examined.
The focal axis of the camera 14 as indicated b~ the
line 24 is aligned with the specular angle for the light
directed onto the surface 18 so that specular light reflected
from the surface 18 is received by the camera. The reflected
.
~L3~0~
light transmitted to the camera is iltered by a polarizing
system 26 which permits only light polarized in the same
direction as the light issuing from the polari~er 16 to pass into
the camera lens system as indicated by the dot in the circle 28.
The beam 11 and the focal axis 24 preferably will be in
a plane substantially perpendicular to the direction of travel of
the web 20.
The weh 20, as indicated in figure 2, is traveling in
the direction of the arrow 30 and the strobe trigger activates
the light 10 to instantaneously illuminate an area such as the
area 32 in Figure 2 with polarized light. The camera 14 is
focussed upon and has a field of view equal to an area 34
indicated by the dash lines and contained within the illuminated
area 32. The operation of the camera 14 and the strobe light 10
are coordinated so that the trigger triggers the strobe light 10
when the camera 14 is in condition to record a frame. The light
10 instantaneously illuminates the area 32 and the surface 18 of
the web 20 is viewed by the camera 14 as a substantially stopped
frame.
The size of the area 34 examined by the camera is
determined by camera optics. Spatial resolution is determined
by the area examined and the number of pixels in the camera pick-
up (e.g. charge coupled device (CCD)). The resolution selected
will normally be in the range of about 0.1 to 0.5 millimeters on
the surface of the web 18 and will depend on the characteristics
being discriminated. It will be apparent that due to the angle
of the focal axis of the camera to the surface of the sheet in
figure 1 to 6 embodiment, the area 34 will be rectangular rather
than square. When sensing printability or surface roughness this
area 34 will normally be at least about 102 cm.
If the resolution is too small the variation in grey
scale level which is formed into a histogram (to be described
herein below) may be unnecessarily wide, in particular the
individual pixels of the camera may register microscopic
variations such as reflection from individual fibres, microscopic
pin holes where there are no fibres, etc. It has been found that
" . ~ :
` ~3~28
for all the instruments described herein, i.e. the surface sensor
and the Formation, blackening and dirt counter to be described
below, resolution generally should not be smaller than 0.1 mm,
otherwise the microstructure Poisson distribution of individual
fiber bundles is examined rather than the macroscale mass
distribution that affects printability.
As above indicated the angle A is preferably
substantially equal to sre~ster's angle for the materia~ being
inspected. Although slight variations of up to about 3 degrees
may be tolerated in the angle it is preferred to be within about
1 of srewster's angle for the particular substrate being
investigated. For paper Brewster"; angle for cellulose has been
found satisfactory.
If desired the instrument composed of the strobe light
10 and camera 14 may be mounted on a suitable carriage 40 which
in turn is mounted on a suitable means 44 for moving the carriage
in the direction of the arrows 42 (generally substantially
perpendicular to the direction of movement 30 of the web). Such
a means 44 may take the form of a threaded rod or the like such
as that illustrated at 44 in Figure 3 which is threaded to the
carriage so that rotation of the rod causes movement of the
carriage axially of the rod 44.
Generally the area 32 is illuminated for a period of up
to about 10 microseconds and preferably for about 3 microseconds
with an intensity to significantly influence the pickup by the
pixels in the camera sensor during the vertical blanking period,
i.e. in the intervals when the pixels are not being read. The
time period or duration of the illumination multiplied by the
speed of the web being examined must be less than the image
resolution if blurring is to be avoided. The CCD or camera pick-
up is not significantly influenced by the ambient light to which
it is otherwise subjected. Not every frame needs to processed,
and thus the strobe need not be triggered for each frame but for
simplicity probably will. The more frequently data is collected
the more information there is available, but also the more
capacity required by the computer. In a conventional camera 1/2
--` 1 3 ~ 2 ~
of the pixels are read durin~ one reading cycle and the remaining
pixels on the next cycle so that any frame grab only obtains half
the information. The output of 1/2 the pixels has been found to
be satisfactory. It has been found that accumulating data and
analyzing the data about every 1 to 2 seconds provides meaningful
information upon which decisions may economically be based.
It is preferred to amplify the analog signal before it
is converted to a digital signal as this improves the quality of
the digitized signal under normal operating conditions.
Furthermore the amplification maintains the analog signals within
accepta~le limits of the analog to digital (A/D) converter in the
digitizer. A simple system for accomplishing this uses a
commercial video digitizer board in conjunction with a personal
computer for example, an AT type personal computer.
A histogram is used for determining printability,
formation and blackening or dirt count as will be described
herein below.
An electronic system suitable for carrying out the
present invention is illustrated in Eigure 4. The strobe light
10 is triggered by a signal pulse taken from the video digitizer
board 47 of the computer 60 and triggers the strobe light 10 in
proper sequence with the video camera 14. The si~nal from the
video camera 14 is carried via a line 46 to the main computer 60
incorporating the digitizer board 47 which may include a buffer
48. The signal is amplified in analog form and then digitized as
indicated at 50. A histogram of the digitized signal is then
produced as indicated at 52, the histogram is analyzed as
indicated at 54, e.g. to determine the standard deviation. The
analyzed frames of data from the camera 14 may then be
accumulated in the accumulator 56 which classifies the data from
each frame. Eventually, based on an accumulation of a suitable
number of frames, (which number may be independently set) the
area of the surface scanned is classified as will be discussed
hereinbelow.
When measuring printability the histogram of each frame
of data as indicated by the curve 62 in figure 5 is a plot of
" ~310~2~
intensity versus frequency (number of pixels) and generally has a
substantially Gaussian distribution for each image area 32. The
standard deviation as indicated by the dimension T in Figure 5 is
determined and used to classify the particular frame as to its
degree of printability. The larger the standard deviation T the
poorer the printability of the sheet. Coefficient of
Printability of the present invention is determined by the
formula:
Standard Deviation
Coefficient of Printability = Mean (1)
Figure 6 shows a plot of MTI print mottle index versus
relative gloss variation as designated by standard deviation (T)
described above. Clearly there is a good correlation between the
data generated using the on-line printability system described
above and the conventional MTI print mottle index.
Formation Tester
A different arrangement of similar equipment is
required to measure Formation. In the arrangement as shown in
the solid lines in Figure 7, the strobe light 100 is positioned
on one side of the paper web 102 and the camera 104 on the
opposite side thereof. As with the previous discussed embodiment
the operation of the strobe light 100 is correlated with that of
the camera 104 and is triggered by a signal from the digitizer
board 106 of the computer 110 so that the camera takes
instantaneous video images of the instantaneously illuminated
areas on the surface of the web 102 in the same manner as
described above for the printability or surface texture sensor.
The frame data from the camera 104 which will examine
an area 34 on the surface of the paper of at least about 102 cm
is amplified, as above described, in analog form and transmitted
to a suitable computer 110 having a digitizer board 106 that may
include a buffer 108, and is digitized by the digitizer 112 to
provide a digitized signal for each frame. Histograms are
11
:
, : :
~ ~31~28
developed in analyzer 114.
It is preferred to normalize the histogram and apply a
gain adjustment factor "F" when it is desired to obtain the
standard deviation as used for determining formation. The gain
adjustment factor F is preferably determined by obtaining the
mean intensity value (M) for each frame of data and then
normalizing the data to a selected mean Ms. Generally the
selected mean Ms will be constant for all frames and will be
selected based on the equipment and the expected intensity
variations. The Factor F is determined by the formula:
F = Ms/M (2)
15 In a particular example given below the value selected
as the normalized mean was MS = 100 units when operating a video
camera digitizer having a range of zero to 255 in intensity.
The standard deviation of the histogram preferably is
then calculated incorporating the factor F and adjusted according
to the following formula:
1 i = 255 1/2
Std. Dev. = (i x F - 100)2 x f~
Total # of Pixels i=0 (3)
25where i = grey level intensity varying from 0-255
fi = # of pixels with intensit~ i
The use of the gain adjustment factor F in determining the
standard deviation further compensates for changes in light
fluctuations in color and/or intensity and further compensates
for basis weight changes.
In determining formation it has been found that the
standard deviation T corresponded well with the Toyoseiki
formation tester total variation value as illustrated in Figure 7
or with the QNSM 'Lin C' value or the inverse of the M/K
formation number. Generally the more concentrated the
distribution, i.e. the smaller the Standard Deviation T, the
better the Formation.
12
'
~3~ l2~
Dirt Counter
The dirt counter is very similar to the formation
tester however the strobe light 100 is replaced by a strobe light
200 positioned above the web 102, i.e. on the same side of the
web 102 as camera 10~ so that light from the strobe directly
illuminates the surface of the web 102 onto which the camera 104
is focussed (dash lines in Fiqure 7). The operation of the
strobe light 200 and of the camera 104 are synchronized so that
the camera 104 generates frames of data when the surface is
instantaneously illuminated by the strobe 200 in the same manner
as described above with respect to the surface texture sensor or
formation sensor. The data for these images is as above described
amplified and transmitted to the main computer 110 into the
buffer 108 and is then digitized to provide a digitized signal as
indicated at 112 and the digitized signal preferably histogrammed
as above described is analyzed in the analyzer 114 particularly
to determine the number of areas or pixels of low reflectance
below a certain selected threshold intensity level as defined by
equation 4.
i = threshold
~ dirt - ~ (f ~ x ~ 100
l = o 1 ~ otal # of pixels J
where i = intensity
fi = no. of pixels with intensity i
In this manner all of the pixels having an intensity
below a preselected threshold level are counted as dirt particles
based on the following:
he threshold for counting dirt is set at any
preselected level but preferably will be a fixed fraction of the
mean intensity level, for example threshold may be equal to the
mean value of the histogram multiplied by a very small number
that may be set by inspection of the segmented binary image to
correspond with dirt specks on a pulp sheet.
The system may then be used to accept or reject
depending on (a) the total area occupied by dirt particles or (b)
13
- . . .
, - '' ~ ~
3~2~
the inclusion of significant large areas of a single dirt spot
or any other suitable criteria that can be distinguished.
This technique may also be used with the strobe on the
opposite side of the paper to the camera, i.e. strobe 100, only
if the web transmits sufficient light.
It is also possible to detect plastic speck
contaminants by substituting a source of infra red light for the
strobe light and using the same analysis technique to determine
the degree and location of plastic elements.
slackening Sensor
The blackening sensor uses essentially the same
configuration and hardware as the formation sensor. To simplify
the operation and accommodate differences in color light
intensity and basis weight, the gain factor F as determined by
formula (2) is employed in this case to modify the selected
blackening threshold ~s5)to correspond with an operating
blackening threshold (s)
' B (operating threshold) = BS (selected blackening threshold (5) F (factor)
and percent blackening is given by the formula
i=255
% blackening = ~ If) x 100
~ l Total # of pixels
i=B
where fi is the number of pixels having an intensity of i
and i = intensity.
Obviously the higher the ~ blackening, the poorer the
paper quality.
All of the described embodiments are suitable for on-
line sensing of selected parameters and may be coupled with other
equipment for example on a paper machine to aid in control.
The preferred embodiments of the invention are
14
i ~31~L28
illustrated by way of examples given herein above and it is to be
understood that the descriptions and drawings are for the
purposes of illustration and as an aid to understanding and not
intended as a definition of the limits of the invention.
Modifications will be evident to those skilled in the
art without department from the spirit of the invention as
defined in the appended claims.
: 15
' ' ,, :, ,
' ~ ' ' : - '
- . ' .
' '
'~
