Note: Descriptions are shown in the official language in which they were submitted.
638
The present invention is directed to inspection of
molded containers, and more particularly to a method and system
for identi~ying a molded container such as a glass bottle or
jar with its associated mold of origin.
Back~round o~ the Invention
Defects in molded containers such as glass bottles
and jars are often related to defects in the associated mold
of origin. For this reason, it is desirable in an automated
operation having a plurality of molds to possess the ability
to identify a specific molded contalner with its mold of origin.
The defective mold may then be shut down for repair while the
remaining molds continue operation. Alternatively, containers
from the defective mold may be automatically sorted as they
proceed down the production line.
Mold identification is generally accomplished by
molding a mold-identifying code into each container during the
forming process. This code may then be read by a suitable
scanner for identifying the container with its mold of origin.
A number of optical techniques, which find particular utility
in conjunction with glass containers, have been proposed for
encoding and later reading of the mold~identifying code. For
example, in U.S. Patent No. 3,745,314, light is transmitted
axially through a stationary container while the image of a
code molded onto the container bottom is rotated past a reading
station positioned beneath the container. Encoding and scanning
systems which are generally similar in basic concept are
illustrated in U.S. Patent Nos. 3,963,918 and 3,991,883, and
in published UK Application Nos. 2,033,120 and 2,017,892. In
63~
general, these techniques rely upon refraction of light passing
through the code bumps or "lenses," and thus on the optical
transmission characteristics of the container material, embody
moving optical elements which do not have desired reliabili~y,
and/or require the use of end codes on the container for
indicating beginning and end of the code to be read. One
embodiment disclosed in U.S. Patent No. 3,963,~18 directs light
from a rotating light source onto a code molded into the bottom
of a container and reads the container code as a function of
light energy reflected from the code bumps.
UK Patent No. 1,580,735 discloses a method and system
for mold identification wherein the mold-identi~ying code is
molded into the heel of the container, i.e. at the junction of
the container side wall and bottom, as a series Oe raised
integral dot or bar bumps disposed in parallel rows or tracks
perpendicular to the container axis. Light energyis transmitted
through the container as the latter is rotated about its axis.
A scanner is positioned to receive light energy transmitted
through the heel of the rotating container. One of the parallel
tracks is treated as a timing track to control reading of code
data represented in the other track, and the code is read as a
function of the refractive characteristic of the timing and
code bumps in the adjacent tracks. Again, start and end codes
are emplo~ed to indicate beginning and end of a scanning
operation.
U.S. Patent Nos. 4,175,236 and 4,230,266 disclose a
mold identification technique wherein the mold-identifying code
i9 molded into the container bottom as concentric rings at
pre~elected spacing. A source of difeused light energy is
~7~38
directed through a gradient filter and a lens onto the bottom
of the container, with the gradient filter being such that the
intensity of light varies linearly as a function of angle of
incidence on the container bottom. A cameral, which includes
an array of photocells parallel to the filter gradient, is
positioned to receive light reflected from the container bottom
as the container passes over the scanning system, and the mold-
specific code is read as a function of the rate of change of
light intensity reflected by the leading and trailing edges of
the rings.
Objects and Summary of the Invention
A general object of the present inventionisto provide
an improved technique for encoding molded containers with a
code specific to the mold of origin and for reading such code,
which technique is economical to implement, reliable in
operation, requires use of only a small portion of the container,
and is insensitive to other indicia such as raised lettering
on the container.
A further and more specific object of the invention
is to provide a method of so encoding a molded container with
indicia identifying the mold of origin, a molded container so
encoded, a method of reading the mold-identifying codes on
containers so encoded, and a system embodying such method.
Yet another object of the invention is to provide a
method and apparatus for optically scanning the container code
which overcomes problems associated with internal reElections
and refractions of light energy within the container wall.
i3~
In accordance with specific em~odiments of the
invention herein disclosed, the containers are encoded during
the molding operation with code indicia consisting of raised
elements i.e. bumps or bars - extending around the heel of
the container in an arcuate array perpendicular to the container
axis. The code elements are arrayed in mutually unique sets
representing binary information which depends upon sequence of
elements within the sets. In the preferred em~odiments, each
set consists of two indicia zones or positions, with the binary
code for each set being a function of the sequence of differing
elements at each indicia position, or the presence of a single
element at one or the other indicia position.
A system and method for reading such code contemplate
a source of semi-diffused light energy directed onto the
container heel as the container is rotated or otherwise moved
in the direction of the indicia array, with the light energy
source having an intensity gradient at predetermined orientation
with respect to the container axis. A camera, which includes
a linear array of light sensitive elements coplanar with the
container axis, is positioned to view the illuminated container
heel. Individual code elements are detected as a function of
light energy reflected by an edge of each element. Following
a complete container scan, consisting of one container
revolution, the data detected by the camera is correlated to
locate the code array, identify individual element sets, and
then identify the code represented by all sets.
Brief Description of the Drawinqs
The invention, together with additional obj~cts,
feature~ and advantages thereof, will be best understood from
7~3~3
the following description, the appended claims and the
accompanying drawings in which:
FIG. l is a schematic diagram of a system for scanning
and reading mold-identifying codes on containers encoded in
accordance with the invention;
FIGS. 2A and 2B illustrate the transmission
characteristics of the gradient filter employed in the system
of FIG. l, FIG. 2A being taken from the direction 2A in FIG. l;
FIGS. 3-6 illustrate respective embodiments of the
encoding scheme of the invention, the scheme of FIG. 6 being
preferred;
FIG. 7 is a magnified view of a code element (FIG.
6) with camera field of view superimposed thereon, which is
useful in explaining operation of the invention;
FIG.~isa graphic illustration of operation ofFIG. 7;
FIG. 9 is a schematic illustration of a normal code
element image as "viewed" by the camera in successive scans;
FIG. lO is a similar schematic illustration of a code
element image which is marginal for detection purposes;
FIG. 11 is a flow chart with accompanying graphic
illustrations showing operation of the scan and read method of
the invention;
FIG. 12 is a fragmentary schematic diagram of a
modification to the system of FIG. l;
FIG. 13 ton sheet 3) is a schematic illustration similar
to those of FIGS. 9 and lO illustrating a code element image which
is marginal for detection purposes; and
~7~38
FIGS~14 and 15 are illustrations respectively similar
to FIG5. 7 and 8, and depict operation of another modified
embodiment of the invention.
Detailed Description
Referring to FIG. 1, a conveyor, typically including
a starwheel (not shown) with a slide plate 20, i5 SO disposed
and connected to a source of molded containers as to bring
successive containers 22 into position at a code read station
24. Conveyor 20 may be of any suitable type, such as those shown
in U.S. Patent Nos. 4,230,219 or
4,378,493 and would typically include a
rotatable starwheel for bringing successive containers
into position and holding the containers in fixed po6ition
during the scanning operation. A drive roller or the like 26
is positioned to engage container 22 at station 24 and to rotate
the container about its central axis 23. An encoder 28 is
coupled to the container rotation mechanism to provide signals
indicative of increments of container rotation. Inthepreferred
implementation of the invention herein disclosed, container 22
comprises a molded glass bottle. A detector 30 such as a limit
switch is positioned to provide a signal indicative of presence
of container 22 at station 24.
A source 32 of semi-diffused light energy is directed
onto the region of the heel 34 of container 22 - i.e., onto
that portion of the container side wall 35 contiguous with the
bottom load-bearing surface 37. Source 32 includes one or more
lamps 36, a diffuser plate 38, a gradient filter 40 and a lens
42 for directing light energy from lamp 36 transmitted through
plate 38 and ilter ~0 onto container heel 3~. Filter 40 (FIGS.
6.
, ~.
,.:~ ~
~i7~8
1 and 2) has a linear attenuation gradient which varies axially
of container 22 (FIGo 2A) from zero attenuation (one hundred
percent transmission) at the upper filter edge to full
attenuation (zero transmission) at the lower edge (FIG. 2B).
A camera 44 is positioned beneath light source 32 and
comprises a plurality of individual light sensitive ele~ents
disposed in a linear array which is coplanar with the axis of
container 22. Preferably, camera 44 includes a linear array
of sixteen elements ~eg., elements 1-16 in FIG. 7). A lens 46
is positioned before camera 42 so as to focus the camera array
onto a field of view which comprises a narrow strip of the
container heel coplanar with and at an acute angle to the
container axis. In general, the angle between the source and
camera is ~aintained as small as possible. A preprocessor 48
receives signals from each of the elements of camera 44, from
container presence detector 30, and from rotation encoder 28
through a divide-by-N counter 49. In general, the purpose of
preprocessor 48 is to monitor the various input signals in real
time, to compress input information into a form indicative of
possible code data, and to transmit such compressed data to a
primary or main processor 50 for analysis. Operation of
preprocessor 48 and main processor 50 will be described in
greater detail hereinafter.
In accordance with the embodiments of the invention
herein disclosed, each container 22 is provided with a code 52
on the container heel 34 for identifying the mold of container
origin. Code 52 comprises a plurality of irregularities, such
as for example, bumps or protrusions 54, integrally molded into
thecontainer heel at uniformly spaced circumferential positions
3~
in an arcuate array perpendicular to the container axis. FIGS.
1 and 3 illustrate an embodiment of the invention wherein each
bump 54 comprises a generally oval bar having a longitudinal
dimension coplanar with the axis 23. Bars 54 are grouped in
sets of two axially offset from each other, i.e., both still
coplanar but at different axial or vertical positions in the
direction of the central axis. Each set is identical to the
remaining sets or the lateral mirror image thereof. Thus, all
bars 54 are grouped in paired sets of two unique configurations,
one of which may be considered to designate a binary "one" and
the other to designate a binary "zero".
Referring to FIG. 3, code 52 is illustrated in the
form of eight paired sets. The right-hand pair set in FIG. 3
is designated by the letter P and is indicative of a binary
parity bit. The succeeding sets, from right to left, are coded
binary bits indicative of the associated 2n decimal values one
through sixty-four respectively. In the particular embodiment
illustrated, a low bar followed by a high bar, moving from right
to left, is interpreted as a binary "one", while a high bar
followed by a low bar is indicative of a binary "zero". Thus,
a binary "one" code at the one, sixteen and thirty-two positions,
and a binary zero at the two,four, eight and sixty-four position,
indicate a decimal mold-identifying code of forty-nine. A "one"
bit in the P position indicates even parity.
FIGS. 4 and 5 illustrate alternative embodiments 52a
and 52b of code 52. The binary and decimal values associated
with the code sets of FIGS. 4 and 5 are identical to -those of
FIG. 3, and FIGS. 4 and 5 are aligned vertically with FIG. 3
for purposes of comparison. In FIG. 4, code 52a comprises a
~ ~7~3~
plurality of long and short bars 56,58 arranged in paired sets
of one long bar and one short bar, different dimensional
characteristics, with the sequence of bars within a given set
being indicative of a binary "one" or "zero". In FIG. 5, code
52b includes a plurality of dot-like bumps 6G arranged in sets
of two and axially offse-t from each other.
FIG. 6 illustrates a preferred embodiment 52c of code
52 in accordance with the invention. FIG.6 is aligned vertically
with FIGS. 3-5 for purposes of comparison. In FIG. 6, the code
52 comprises a circumferential array of irregularities or dot-
like bumps 62 similar to the bumps 60 in FIG. 5. However, in
the embodiment of FIG. 6, the bumps 62 are not grouped in paired
sets to be axially offset or dimensionally different from each
other as in FIGS. 3-5. Instead, each set has only one bump 62
at either one of the circumferential positions laterally aligned
and paired with an empty space at the other circumferential
position, as illustrated by the phantom representation 64 in
each set. Since, in the embodiment of FIG. 6, the sets or code
bits are represented by a bump 62 and an empty space 64, rather
than by two dots or bars as in FIGS. 3-5, it is necessary to
provide start and end code bits. Thus, the embodiment of FIG.
6 does not include a sixty-four bit or a parity bit, although
such could be added by merely lengthening the code in the
circumferential direction.
Operation of the invention for scanning and reading
the mold-identifying code on each bottle will be described in
connection with FIGS. 7-11 and 13. In general, the individual
surface irregularities which collectively comprise the cod~,
specifically bumps 62 of code 52c (FIG. 6), are detected as a
~ ~i7~3~
function of disturbance or variation of the normal pattern of
light reflection from the container heel as the camera "sweeps"
the code - i.e., as the code is moved past the stationary camera
and light source. Normally, the individual camera photocells
view a generally gray field at the intensity of the mid-line
41 (FIGS. 2A and 2B) of gradient filter 40 (FIG. 1). However,
the upper and lower sloping surfaces of each bump 62 re~lect
the field of view of one (or more) photocells upwardly or
downwardly of filter 40, i.e. in the direction of the filter
gradient, so that the intensity or brightness "seen" by that
photocell is greater or less than that "seen" by adjacent
photocells whose field of view is not so reflected. The system
electronics scan the photocell array (elements 1-16 in FIG. 7)
of camera 44 at increments of containerrotation(scan increments
in FIG. 9), identify the code elements or bumps 6? as "bright
spots" in the swept field, and thereby determine the code
represented by the code elements and the spacing therebetween.
FIG. 7 illustrates the field of view of the sixteen
photocells of camera 44 superimposed upon a code bump 62 at a
scan which centers upon -the code bump. FIG. 8 illustrates the
light intensity or brightness pattern seen by camera photocells
7-14 for the specific scan of FIG. 7, and FIG. 9 illustrates
the overall image 62a of bump 62 seen by camera 44 as the camera
sweeps the bump. Preprocessor 48 (FIG. 1) sequentially scans
or sweeps each of the light sensitive photocells 1-16 in camera
44 at each increment of container rotation sensed by encoder
28 and counter 4g. When no code indicia bump is within the
camera field of view, all camera elemants sense a gray background
intensity I (FIGS. 8 and 9). However, when a code indicia bump
10 .
62 moves into the camera field of view, the slope of the bump
top edge disturbs this normal reflectance pattern, intensifying
light energy seen at some elements while reducing the intensity
seen at other elements. In the particular scan of FIGS. 7 and
8, the upper slope of bump 62 reflects the field of view of
photocell 11 relatively high on the gradient Eilter so that the
intensity at photocell 11 is highest for that particular scan,
with the intensities at photocells 9-10 and 12-14 being greater
than I (gray) but less than that of photocell 11. In the same
manner, fields of view at photocells 7 and ~ will be reflected
down~ardly by the lower edge of bump 62 to a darker portion of
the gradient filter, so that the intensities at photocells 7 and
8 are less than gray intensity I.
The resolution of camera 44 depends in part on the
width of indicia irregularities in increments of container
rotation, i.e. the number of camera scans per unit width, the
distance D (FIG. 6) between indicia positions, and the axial
length of the indicia irregularities in relation to the number
of camera elements. In a preferred embodiment of the invention
(FIG. 6) wherein the indicia irregularities comprise circular
bumps, the diameter of such bumps (i.e. length and width) is
about 0.080 inches (FIG. 7). The distance D between indicia
position is 0.104 inches, and the scan increments (FIG. 9) are
equal to 0.0065 inches, yielding sixteen scans between adjacent
indicia position centers and twelve scans across each bump 62.
Effective separation S (FIG. 7) between photocells at the
container surface is 0.013 inches.
Resolution also depends upon the angle of light source
32 and camera 44 to bumps 62 and thus to the container axis.
For example, if the light source and camera are positioned too
low, it is possible that the upper slope of the bump will reflect
the ield of view of one or more camera photocells over the
gradient filter rather than to a bright upper portion thereof.
In such an e~entuality, the bump image would appear as in FIG~
10, with a dark region 62c where the brightest portion of the
image should appear. Preferably, light source 32 and camera
44 are made adjustable as an assembly for empirical adjustment
to obtain an acceptable image. Another marginal detection
condition is illustrated in FIG. 13, wherein the dot image is
generally dark with only a small bright spot 62d just above
center. An image of this type, which is still detectable, may
result from an accumulation of material in the container mold
so that the dot bumps are not as sharp or pronounced as desired.
Operation of preprocessor 48 and main processor 50
(FIG. 1) will be best understood with reference to FIG. 11.
Thecode52c illustra-ted in FIG.ll isthe same asthat illustrated
in FIG. 6. The nominal arcuate length L between bumps 62 at
the beginning andend of the start and end codes is predetermined.
Likewise, the nominal distance D between indicia positions is
known. Preprocessor 48 identifies thecamera photocell receiving
peak light intensity in each camera scan. This is accomplished
by comparing each cell intensity with that of a selected lower
cell, preferably the fourth lower cell, and identifying a peak
intensity change. That is, the intensity at cell 5 is compared
with that at cell 1, the intensity at cell 6 is compared with
that at cell 2, and so on. The differential of four is selected
as a function of bump size. Thus, in the scan of FIG5. 7 and ~,
the peak differential is associated with cell 11 (i.e., cell
12.
i7~3~3
11 compared with cell 7). If this differential is greater than
an empirically selected threshold Tl, the absolute value of
such peak dif~erential, together with the associated cell number
(11) and the scan number, is transmitted to main processor 50.
This process is repeated at each scan increment for an entire
container rotation. Thus, at the end of one rotation, there
is stored in the memory of processor 50 a block of data
representing peak differential intensity data (above
differential threshold Tl), and scan and cell numbers associated
with such peak data. The task of main processor 50 is to
distinguish a legitimate code from other irregularities, such
as raised letters 66 (FIG. 1) on the container heel, and to
read such code.
This is accomplished in main processor 50 by first
finding blocks of data having an arcuate length and intensity
pattern which might indicate a legitimate code, i.e., CODE FIND
(FIG. 11). A moving average MAl is obtained by summing at each
scan increment all peak intensity differentials for a group of
4D adjacent scans. The moving average MAl for each data block
(only the legitimate data block is illustrated in FIG. 11) is
then analyzed to find one (or more) which has an amplitude
greater than a preselected threshold T2 for a duration between
0.9L and l.lL, with no dips 68 below threshold T2 of more than
2D duration. All peak intensity differential data blocks which
satisfy these criteria are analyzed further to find individual
code dots.
To locate individualcode dots,a second moving average
MA2 is obtained by again summing peak intensity difEerential
data at each scan increment, this time over a group of ~D
3~
adjacent scans, i.e., DOT FIND (FIG. 11). Because of the
narrower summing range ~i.e., ~D compared with 4D), code dots
or apparent dots will appear as peaks 69 above a "gray" background
intensity 71. Start and end data bumps are then preliminarily
identified at peaks 69a969b, and moving average MA2 i8 scanned
in increments of D to identify presence or absence of a dot at
each indicia position. (It will be recalled that potential dot
positions are of known uniform spacing D, so that peaks 69
indicative of potential dot-code information should be at
increments of D between start and end peaks 69a~69b.) Any dot
data so located is then paired to find code bits, i.e., BIT
FIND (FIG. 11) which identifies the binary code indicative of
mold of origin.
Most preferably, for increased reliability, the CODE
FIND, DOT FIND and BIT FIND steps of FIG. 11 are carried out in
an iterative process, with an ACCURACY FIGURE being determined
for each data block that might represent a legitimate code, and
the data block having the highest ACCURACY FIGURE being
identified as the dot code. This ACCURACY FIGURE is obtained
by examining the second moving average MA2 for each data block
after the BIT FIND step to find the bit pair having the smallest
amplitude differential. In the example of FIG. 11, the smallest
differential is illustrated at 70. Such minimum differentials
for each data block are compared, and the data block having the
highest minimum bit pairdifferentialis identified asindicating
the dot code. A similar process may be employed for the code
embodiments of FIGS. 3-5.
As previously indicated, disposition o gradient
filter 40 ~FIG 1) such that transmissivity increases upwardly
14.
~i76;~
functions in accordance with the invention to detect the upper
slope of the dot or bar protrusions. Thus, all of the embodiments
of FIGS. 3 5 would appear substantially identically to the
camera and detection e]ectronics. The embodiments of FIGS. 3
and 5 could also be read with the transmissivi~:y of the gradient
filter increasing in the downward direction so that the system
would detect lower slope of the protrusions. The embodiment
of FIG. 4 could not be so read since the lower slopes are
aligned, so that long and short bars would appear the same to
the camera. Of course, the embodiment of FI~. 6 could be read
either way.
FIG. 12 illustrates a modification 78 whereby the
system detects and is responsive to the side edges or slopes of
the protrusions. Light source 80 and camera 82 are positioned
side-by-side, and the gradient filter 84 of light source 78 has
a transmissivity characteristic which varies laterally - i.e.
in the direction of container motion - rather than in the
direction of the container axis. The array of photocells in
camera82 iscoplanar withthe container axis as in the embodiment
of FIG. 1 so as to extend across the code area on the container
heel. The "image" of dots 62 "seen" by camera 82 will be the
same as that illustrated in FIG. 9 rotated 90 cloc~wise. Thus,
the orientations of the gradient filter and lamp/camera
combination determines the protrusion edge which will be
detected. The embodiment of FIG. 12 is preferred for amber
glass. The embodiment of FIG. 1 is preEerred for flint glass
where internal reflections may cause a "hot spot" problem with
the embodiment of FIG. 1~. The invention is also useful for
reading codes on non-circular containers such as Elas~s or
3~
square bottles. The code may be molded onto the heel beneath
a substantially flat container wall and read by moving the
container linearly past a read station.
The embodiment of the inventionhereinabovediscussed,
wherein the intensity at each camera cell is compared with the
intensity at the fourth lower cell in each scan, is preferred
in theory. However, it has been found the elements 1-16 of
differing cameras 44 are not always sufficiently closely matched
to yield reliable data. Thus, for example, if the temperature
characteristic of cell 11 differs from that of cell 15, a peak
differential may be indicated at cell 15 even though no such
differential has in fact occurred. When the camera cells are
not sufficiently closely matched, the technique of the invention
may still be implemented in the manner illustrated in FIGS. 14
and 15. More specifically, intensity data is stored by cell
in preprocessor 48 for four consecutive scans, with the oldest
scan data being discarded when each new scan is taken. The
intensity at each cell in a given scan is then compared with
the intensity at the same cell in the fourth preceding scan.
Thus, in the illustration of FIGS. 14-15, the intensity at cell
9 at scan N is compared with the intensity at the same cell at
scan N-4, and the peak differential (FIG. 15) is thus identified.
The remainder of the process - i.e., CODE FIND, DOT FIND and
BIT FIND, remain as previously described.
A disadvantage of the code embodiment of FIG. 6 as
compared with those of FIGS. 3-5 isthat,because of the necessity
of START and END bits, a lesser maximum code value is possible
for a given number of indicia positions (sixteen in all FIGS.
3-6). This limitation can be overcome to some extent by grouping
~tj7t~38
the indicia positions in sets of more than two. For example,
in the embod.iments thus far discussed, each set o:E two lndicia
positions can have either of two binary values (O or 1), and
two sets of four adjacent positions can have a total of four
possible binary values (00, 01, 10 or 11). If the indicia
positions are grouped in sets of four, with the limitation that
each set must contain two dots (D) and two blank spaces (B~,
the following six comb.inations are possible:
B B D D
B D B D
B D D B
D B B D
D B D B
D D B B.
Summary of the Disclosure
Thus, in accordance with a first aspect of the
disclosure and invention, a molded container, preferably a
molded glasscontainer, is provided with coded indiciaindicative
of the mold of origin in the form of a plurality of irregularities
extending in an array along a preselected location on the
container outer surface, preferably in a circumferential array
around the container heel. In each of the embodiments of FIGS.
3-6, the irregularities comprise outwardly projecting bumps or
protrusions disposed at selected indicia positions, such
positions being uniformly spaced around the heel. In each
embodiment, the indicia are provided in two mutually unique
sets at sequential pairs of positions, the two sets being the
lateral mirror image of each other and thus being adapted to
17.
~L2~i7~3~
represent binary characters which collectively comprise a binary
code indicative of the mold of origin.
In the embodiment of FIGS. 3-5, a bump or protrusion
is provided at each indicia position, with the two bumps or
protrusions of each set having differingdimension or positioning
in the direction of the container axissoasto bedistinguishable
from each other~ In FIGS. 3 and 5, the protrusions are identical
and are offset in each set axially of the container. In ~IG. 4,
the protrusions are non-identical within each set. The
embodiments of FIGS. 3-5 have the advantage of not requiring
start and end bits at the ends of the code. In the preferred
embodiment of FIG. 6, a protrusion is provided at only one of
the two indicia positions within each paired-position set, the
other position being "blank". The protrusions are aligned
circumEerentially of the container axis and are all identical.
The embodiment of FIG. 6 has the advantage of requiring less
memory within the processing electronics.
In accordance with a second aspect of the disclosure
and invention, a system and method are provided for scanning
and identifying the container code. The system includes a
source of diffused light directed onto the container heel and
having a light intensity gradient which varies ina predetermined
direction with respect to the container axis. For flint glass,
it is preferable that the intensity gradient of the light source
vary in the direction of the container axis so that the system
functions to detect upper slopes of the code bumps, and so that
light energy reflected from internal surfaces is directed out
of the camera field of view. For amber glass, it is preferred
18.
7~3~3
that the gradient be lateral to the container axis so that the
system is responsive to side slopes of the code bumps.
A camera is positioned to receive light energy
reflected by the container heel. The camera includes an array
of light sensitive elements, preferably sixteen, perpendicular
to the code array and, thus, coplanar to the container axis
where the code extends circumferentially around the container
heel. The container code is moved past the container field of
view, by rotating the container in the case of the usual
cylindrical container or by linear motion where the code is on
the flat portion of a flask-type container. A preprocessor
scans the camera array at increments of container motion.
In accordance with the method of the invention
described in detail in connection with FIG. 11, peak illuminance
data is transmitted at scan increments by the preprocessor to
a main processor. In the main processor, blocks of peak data
are analyzed in an iterative process to first locate data blocks
which might represent the container code, then to identify
individual code bumps or irregularities, and then to associate
such irregularities by pairs of indicia positions so as to
identify possible code bits. These code bits are then analyzed
to identify the data block most likely to represent an actual
code, and the actual code is then determined.
19.
