 |
Description  |
|
|
BACKGROUND OF THE INVENTION
In the field of machine-readable information code, the bar code is one of
the most widely recognized by reason of being used on numerous consumer
articles and in other applications. One bar code, being the universal
product code (UPC), includes different width bars which are spaced one
from another in an overall pattern on a consumer article for scanning
along a path of travel by a moving reader or sensor. Contrariwise, the
path of travel of a consumer article carrier with respect to a reader or
sensor could be used wherein the reader or sensor is maintained stationary
for scanning articles moving therepast. The different width vertical bars
and the spaces therebetween make up the UPC scannable symbol, and optical
font characters make up the corresponding human-readable characters. The
vertical bars and spaces reflect light emitted by an optical scanner which
reads the symbol-marked products--such UPC symbol identifying the
manufacturer and the product.
Machine readable codes may also be triangular-shaped of equal length bars
or they may be of unequal length wherein the sensor identifies characters
by reading along the triangles or across the various lengths thereof.
Another form of bar code is the color bar code which may consist of black,
green, and white bars arranged in various patterns for representing
characters.
A further type of code includes the optical character recognition type
(OCR) which may be both machine and human-readable and is utilized where
both readings are desirable.
Another code pattern utilizes a matrix which may consist of squares, dots,
circles or like indicia which is machine-readable and which is useful for
identifying and printing of characters.
The above codes are generally formed along a line wherein the reader or
sensor follows a straight path across the code wherein the code is
directed and disposed in a path along a line coinciding with the direction
of movement and viewing of the sensor. In most prior art codes, the entire
pattern usually must be sensed and the relation between the pattern and
the sensor usually must be fixed to properly recognize the characters,
however, the hand held wand readers allow some variation from a fixed path
or line of travel.
The basic concept underlying recording is the creation of patterns in or on
a record medium so that one or more patterns can be taken to represent one
or more characters. The patterns are sequentially detected by appropriate
sensor arrays when there is relative motion between the record medium and
the array.
Many applications exist for the several printed codes mentioned above,
which codes have high information density and are machine readable. While
none of such codes can be used for all applications, the following
parameters or requirements specify a particular code for very wide usage.
The code should be high density for printing of at least ten alphanumeric
characters per linear inch, the code should be readable with a light
weight, low cost, and preferably hand-held wand, and the first pass read
rate should be at least or greater than 95%. The character substitution
error rate should be less than one in ten million after all error
detection and error correction has taken place, and the code should be
printable with a low cost, computer controlled, widely available printer
such as a dot matrix printer.
A dot matrix printer is defined as one capable of printing dots on a record
medium at selected points on an X-Y grid and representative types of such
printers include a wire or needle matrix impact printer, a thermal
non-impact printer, or an ink jet non-impact printer. Some of these
printers utilize dual grids in which characters are located on a coarse
grid having center-to-center spacing 0.1 inch horizontal and 0.16 inch
vertical, while within a character, dots are located on a fine grid with
both horizontal and vertical spacing of 0.015 inch with typical characters
printed at seven dots high and five dots wide.
In many cases of the prior art, when the reading depends essentially on
multiple sensing elements, it is seen that the code scheme requires that
the reader or sensor array have elements in an order where any one element
always senses the same portion of the particular pattern, or stated
differently, the allowable unwanted lateral displacement of the sensor
with respect to the record medium must be kept less than the lateral
dimension of a sensing element--the lateral dimension being along a
direction perpendicular to the relative motion of the medium and the
sensor. In specific applications, for example, where the reader and the
symbols are under machine control, the reader (read head) must read the
same track on magnetic tape, or the photocell or contact must read the
same bit of each character on paper tape. That is, the sensor array must
be properly aligned with respect to the pattern so that there is no
ambiguity about the meaning of the signal coming from each element of the
array.
Representative prior art which is considered relevant to the subject matter
of the present invention includes U.S. Pat. No. 3,532,859, issued to J.
Laplume on Oct. 6, 1970, which discloses an identifying system using
optical codes wherein a plate or card comprises a plurality of squares in
characteristic binary code patterns of non-reflective and reflective
surfaces. The reflective surfaces may comprise a multiplicity of small
spherical beads or a multiplicity of convex elements alongside the
non-reflective or light-absorbing surfaces and be read or sensed by an
illuminating and sensing unit.
U.S. Pat. No. 3,558,859, issued to F. W. Dilsner et al. on Jan. 26, 1971,
shows an automatic reading system for record media having encoded data of
perforated and of printed codes. The perforated code includes legible and
illegible type arranged in a matrix pattern and the printed code includes
a bar code and a marking code.
U.S. Pat. No. 3,860,790, issued to S. J. Reckdahl on Jan. 14, 1975, shows a
data processing form which has a plurality of printed indication areas for
use with optical sensing apparatus. Printed symbols extend uniformly
within a certain dimension and have a printed area of an optical density
which is selected in relation to the sensing means.
U.S. Pat. No. 3,898,434, issued to A. G. Bigelow et al. on Aug. 5, 1975,
discloses a machine-readable coded member formed of pattern areas with
each area corresponding to a character. Each area is divided equally into
rectangular portions and the presence or absence of indicia in selected
portions represents a designated character. At least two pattern portions
of each area are aligned along a path of travel of the coded member with
respect to the scanner or reader and an index mark is placed on a line
which is perpendicular to the path of travel of the document.
U.S. Pat. No. 4,114,033, issued to A. Okamoto et al. on Sept. 12, 1978,
discloses a number of bar codes recorded in dual directions on an
information card. A plurality of bar codes comprising one group bar code
are arranged in one direction which is orthogonal to the bar symbol and a
plurality of group bar codes are arranged in the other direction in which
the bar symbol extends. The group bar codes are sequentially scanned for
the recognition thereof.
And, U.S. Pat. No. 4,130,243, issued to R. L. Stevens on Dec. 19, 1978,
discloses a machine-readable optical printed symbol format that is
generally hour-glass shaped and consists of nine elements of alternate
bars and spaces vertically arranged in a manner wherein the length of the
bars and the spaces is greater at the top and bottom of the symbol and
decreases to a minimum point at the middle of the symbol.
SUMMARY OF THE INVENTION
The present invention relates to machine-readable codes and more
particularly to a high-density code of the dot matrix type. The code
consists of a plurality of dot patterns in columnar fashion wherein each
of the patterns in a column represents a character with the dot patterns
being preferably repeated in one direction. The dot matrix patterns are
read by a reader or sensor in a direction of reading which is the X or
abscissa direction and which is normal to the direction of the repeated as
well as non-repeated patterns which are in the Y or ordinate direction.
The code is constructed and oriented in a first or preferred manner
wherein the reading or identification of each of the matrix patterns may
be accomplished with a reader having a field of view which covers a
portion of the entire code area or in a second manner with a reader which
views a non-repeated code and specifically which includes and covers an
area in excess of one matrix pattern in the Y direction.
The code is made up of dots spaced from each other in the X and the Y
directions, with the dot columns representing characters digitally in the
manner of regularly spaced parallel columns of dots and with various
sequences of fully-populated columns and empty columns indicating the
start and finish of a symbol or of a plurality of characters. The
character code extends in the Y direction and the binary information is
conveyed by sensing or reading the presence or absence of dots at specific
places or locations within the dot columns and within the matrix pattern
for each character or characters.
In view of the above discussion, the principal object of the present
invention is to provide a machine-readable matrix code in a simple
configuration.
Another object of the present invention is to provide a matrix code of high
density for conveying binary information.
An additional object of the present invention is to provide a matrix code
in a repeated pattern wherein reading means can be randomly moved across
the code within a given range in one direction.
A further object of the present invention is to provide a matrix code in
repeated patterns in one direction to provide for correction of tracking
error of the code-reading means.
Another object of the present invention is to provide a code which can
readily be printed by dot matrix printers commonly used to print computer
output data.
A further object of the present invention is to provide a matrix code with
inherent error detection features.
Additional advantages and features of the present invention will become
apparent and fully understood from a reading of the following description
taken with the annexed drawing.
BRIEF DESCRIPTION OF THE DRAWING
FIGS. 1A and 1B represent a plan view of a matrix code with repeated
patterns in a preferred embodiment of the present invention, together with
a reader for reading the code;
FIGS. 2A and 2B represent a plan view of a matrix code and a reader having
a field of view for sensing the code;
FIGS. 3A and 3B represent a plan view of the matrix code shown in FIGS. 2A
and 2B and depicting a particular application thereof, along with a reader
for sensing the code;
FIGS. 4A and 4B represent a plan view of the matrix code shown in FIGS. 1A
and 1B and depicting a particular application thereof, along with a reader
for sensing the code; and
FIG. 5 is the same view as FIG. 4A with the code shown in evenly spaced dot
matrix manner.
DESCRIPTION OF THE PREFERRED EMBODIMENT
Prior to describing the several figures, it should be stated that the
fundamental concept of the present invention is to represent characters
digitally by regularly and evenly spaced parallel columns of dots and
where binary information is to be conveyed by presence or absence of a dot
at any specific location on the record medium or paper. The possible
locations for dots will be spaced along the columns so that the minimum
dot-to-dot distance in one column is the same as the distance between
columns and the distance between rows of dots, although other column and
row spacings may find useful application. Thus the possible locations for
dots form a regular array, grid or matrix with the dot columns running
vertically and the dot rows running horizontally.
Detection of the dots is to be performed by an optical sensor or reader
which has a similar, but possibly dimensionally different, grid structure,
one axis of which is aligned with the code grid within a few degrees. As a
special case, the grid structure of the optical sensor may be a single
column of sense elements arrayed approximately in the vertical or Y
direction. If the sensor is in a hand-held wand, the mechanical,
electronic and digital systems must correct for or prevent errors arising
from non-ideal orientation and motion of the wand. In such a wand or like
hand-held reading device, accurate alignment is extremely difficult or
almost impossible to achieve and, generally, any one element of the sensor
array scans a different part of each code pattern each and every time the
device makes another sweep across the record medium. The system must be
capable, within limits of course, of making sense of and identifying the
data regardless of what part of the sensor array passes over any
particular part of the character pattern. Additionally, dot matrix
printers are not perfect in their certainty of laying down dots of
required optical contrast and spurious "dots" may appear which are caused
by dirt or by defects in the record medium or paper.
The matrix code refers to the representation of one character by a single
column of marks regularly spaced in a line, one class of mark representing
binary one and a second class of mark representing binary zero. A black
mark can represent binary one and a white mark or absence of black can
represent binary zero, with the array of printed rows and columns of dots
being a symbol and the overall invention being identified as a symbolic
coding method.
In an arrangement of vertical dot columns of code, angular tracking error
or drift may be corrected by repeating the n bits of the code for one
character several times in a single column. The vertical height of the
sensor or reader in the plane of the paper will be such that at least one
full n-bit code height is detected regardless of the position of the
sensor along the column of dots. Codes which can be converted into one
another by cyclic permutation will be considered equivalent and will
represent the same character or data. Cyclic permutation indicates or
signifies that any bit may be the start bit when the code is arranged in
an imaginary circle of exactly n bits around the circumference, such
codes, which can be uniquely recognized without reference to a particular
start bit, will be referred to as cyclic patterns or codes. Thus if the
code is repeated several times in a vertical column by joining start to
end bits, the reader or sensor can select any n bits from the vertical
column of dots and uniquely identify the code. In this manner the sensor
is narrower than the repeated vertical rows of dots and the code
represents a redundant pattern.
Cyclic code patterns have inherent error detection capability of two types.
First, since the machine will reject any code which is not a cyclic
pattern and since the cyclic patterns are a small fraction of all possible
patterns having the same number of bits, errors consisting of unwanted
marks on the medium or of missing dots have a low probability of changing
a valid cyclic pattern into another acceptable pattern. Such errors will
thus be detected and an alert can be given to the operator. Second, since
the reader is larger in vertical extent than the length of one complete
n-bit pattern, the machine can recognize more than one contiguous group of
n dots and spaces to be decoded. All of these groups in any one column can
be decoded during the reading period and must correspond to the same
character. If not, the machine will signal an error.
Velocity error is corrected by including repeated rows of dots in a
vertical direction in the printed symbol which are not recognized as code
but as timing or fiduciary marks. As an example, a three-row combination
of all white, all black, and all white marks could easily be sensed by a
sensor or wand. The rate of occurrence of such marks would be analyzed to
give probe velocity and thus used to generate a data clock. Further, the
inclination or skew of the probe can be deduced from the sequence of times
at which different sense elements of the probe pass over any one timing
row. All of the data in the following codes would be corrected for the
timing errors introduced by the slant or skew of the reader on the
assumption that the angle was changing slowly.
As an example, if a code is represented by 11 dots on 0.015 inch centers
repeated four times in a vertical direction, the symbol is 0.66 inch high
and the sensor field needs to be approximately 0.20 inch to see the 11
significant dots and have one or two guard dots on each end. The 11 cyclic
dots can be arrayed in 188 unique ways and can represent a complete upper
and lower case alpha font with numerics and symbols.
It is possible to provide 66 vertical columns of such dots per horizontal
inch and if three columns of each set of 11 columns are used for timing
marks, the symbol will represent 48 characters per horizontal inch. If the
symbols are spaced on one inch centers vertically and 7 inches of the
paper width is used, 70 inches of symbol or 3360 characters per 81/2 inch
by 11 inch page can be printed, which is about equal to the number of
characters on a single spaced typewritten page of human readable print.
In the preferred embodiment of the invention, the reader or sensor field of
view is less than the length of a repeated vertical column of dots. The
column of dots may contain as many as 50 dots but any 11 dots in sequence
can be decoded to represent the character in question regardless of which
dot is taken as the first bit of the code. If, for example, the reader or
sensor is wide enough to always detect 13 dots reliably, the logic can
pick out a sequence of 11 dots and uniquely assign the proper character.
In this manner, the operator of the sensor can drift from top to bottom or
from bottom to top of the repeated symbol without making errors as the
symbol is scanned from left to right or from right to left all within the
capabilities of the system.
A second embodiment or an alternate approach, especially applicable if
limited fonts are adequate, is to print codes having the same number of
bits as the number of wires in a conventional print head, namely 7, 8 or 9
print wires, and which can print dots simultaneously in a vertical
direction. In a typical 7 wire configuration, six dots per code can be
used for a 64 character set with one dot reserved for the start position.
Additional dots can be used for error checking.
Symbols made up of such codes would be printed in a single pass of the
print head and each vertical column of dots would contain a single normal
binary code without cyclic repetition. The sensor field would need to
extend far enough above and below the printed symbol to allow for normal
drift during the sweep of the sensor or wand reader.
The second scheme is simpler than the first scheme by reason of the code
dots having a one-to-one correspondence with the bits of a simple binary
code, such as ASCII. The field of view of the sensor is sufficiently wide
to overlap the code symbol above and below into regions of the record
medium where there are no printed dots. The extra width of the sensor is
sufficient to allow for operator drift while sweeping or scanning the
code.
There is an implication in all suggested code systems that allows for
misalignment between sensor and code, which is that errors must not arise
when the field of view of one sense element overlaps two pattern elements.
One solution is to use a fine "grained" sensor so that each pattern
element will always encompass the full field of view of at least one
sensor element. The system logic then decides which sensor elements are
"pure" in that they convey the signal from only one pattern element, and
which sensor elements are "mixed" must either be corrected or ignored. It
is believed that, when applied to the present invention, the
center-to-center spacing of view fields of the sense elements must be less
than or equal to one-half the minimum center-to-center spacing of printed
dots.
The choice between symbols with repeated codes to be read by a sensor
having a field of view shorter than the code column of dots and symbols
with non-repeated codes to be read by a sensor having a field of view
longer than the code column of dots depends upon the cost of the sensor
and the decoding hardware for the two embodiments. The ease of following
the code during hand sweeping of the wand reader also is a determining
factor of which code to use.
In the repeated or redundant pattern case, the reader or sensor sees only a
part of the pattern but is capable of decoding such pattern. In the
non-repeated or overlap case, any part of the sensor is capable of
completely decoding the pattern as long as the entire pattern is covered.
As mentioned previously, matrix printers selectively deposit dots on the
record medium or paper at locations which specify a regularly spaced grid
and the presence or absence of such dots at the spaced locations along one
column of the grid represent the bits of an n-bit binary number or code.
The binary data is repeated several times along each column and only those
binary codes are used which can be uniquely recognized in any cyclic order
without reference to most and least significant bits. The printed codes
are read with the optical reader or sensor which has at least n optical
elements arrayed in a line approximately parallel (within the precision of
hand alignment) to the columns of the code. The sensor is moved in a
direction approximately perpendicular to the code columns so as to detect
the sequence of codes.
Referring now to the several Figures of the drawing, FIGS. 1A and 1B
illustrate a preferred embodiment of a dot matrix code 10 of the present
invention wherein such code comprises rows 12 of dots in the X direction
and columns 14 of dots in the Y direction. The particular code illustrated
shows a combination of dots and spaces totaling nine and arranged in the Y
direction to comprise a 9-bit code for each character. Reading or counting
from the top of FIG. 1A and in the case of the letter "A," the code has
eight vertical dots and a space, the letter "B" has seven vertical dots
and two spaces and the letter "C" has four vertical dots, a space, a
single dot, a space and two dots. The code for letter "C" could equally
well be considered to cover six dots, a space, a single dot and a space.
The presence of dots or the absence of dots make up the matrix code for
the respective letters and numerals. A 9-bit code is usually taken to
represent a maximum of 2.sup.9 or 512 possible different characters.
However, only 58 of these characters, excluding full and empty columns,
are unique in cyclic form.
It is readily seen from the dot matrix codes of FIGS. 1A and 1B that the
9-bit code for each letter or numeral is repeated twice in the Y direction
to form the redundant pattern. The spacing of the dots is arbitrary and is
employed specifically for convenience in showing the dots separated from
each other in the Y direction for ease of illustration and for permitting
adequate space for showing the letters and numerals in the X direction.
For example, the spacing or distance between dots, as represented by "a"
and by "b" may be reduced to zero so that the adjacent dots are touching,
as can be accomplished where the codes are printed by a dot matrix printer
of any one of the several kinds as mentioned above. A common matrix
printer may have almost any desired dot spacing in the X direction and
with little modification any desired dot spacing in the Y direction. A
reader 16 is shown at the left side of FIG. 1A for reading the dot matrix
code, which reader, for example, may be a wand-type reader as manufactured
by Caere Corporation, of Mountain View, Calif. The reader 16 has a field
of view sufficiently wide to cover more than nine dots and/or spaces in
the Y direction so as to always see a full 9-bit code, regardless of how
the wand is positioned vertically within the code area. As long as the
reader is moved along a path through the repeated dot matrix code, the
character represented by the vertical column of dots and spaces is sensed
or read and retrieved for future use. The reader 16 can move in a slanted
or skewed or skewed manner across the code pattern, as seen in FIG. 1B,
wherein it is well-known that a hand-held wand reader does not always
travel along a precise line or plane when reading the code. In this
respect, the logic of the control system is intended to correct for the
skew of the reader.
FIGS. 2A and 2B illustrate an overlap pattern of a dot matrix code 20
wherein the code may be printed by means of a seven element dot matrix
printer (not shown) in the printing of the seven dot high code element.
The codes for the respective letters and numerals are illustrated as the
presence or absence of dots in a lesser height of the code. A reader 22
must overlap the top and bottom of the code symbol so that the field of
view of the reader includes the height of the 7-bit code. Again, as
illustrated in FIG. 2B, the reader 22 may be skewed in its travel along
the code and the logic will correct for this condition. A fully populated
column 24 of dots at the start of a symbol and a blank column 26 adjacent
thereto indicate both start and finish or end of a symbol. Various
sequences of full and blank columns are used to signify start and finish
of a symbol or division of the symbol into blocks of data, e.g. a pair of
full columns, a blank column and a single or a double column of dots. The
full and empty columns 24, 26 may divide blocks of code and the spacing of
the columns is used to determine wand speed to assist in decoding. The
repeated double dots 28 indicate end of transmission. A column containing
a single dot 29 indicates a space or a blank character.
FIGS. 3A and 3B illustrate a particular application of the overlap code 30
in spelling out a "37 character ASCII subset" by reading or sensing
thereof by a reader 32. A full column 34 of dots and an adjacent blank
column 36 in a pattern or sequence (FIGS. 3A and 3B) indicate start and
finish of the symbol. The repeated double dots 38 indicate end of
transmission and a column containing a single dot 39 is a space or a blank
character.
FIGS. 4A and 4B illustrate a particular application of the redundant
pattern code 40 in spelling out a "58-character alpha numeric dot code"
together with a reader 42. In similar manner, a fully populated column 44
of dots and an adjacent blank column 46 indicate start and finish of the
symbol. The single dot 48 is end of transmission and the column of dots 49
is a space or blank character of the symbol.
FIG. 5 represents the identical code as FIG. 4A with the code shown in
evenly spaced dot matrix manner.
FIGS. 2A, 2B, 3A, 3B, 4A and 4B illustrate start and finish of the symbol
by use of a pair of fully-populated columns, a blank column, and another
pair of fully-populated dot columns at the left side of the code for
indicating start of code. The right side of the code shows a single
fully-populated dot column, a blank column and a pair of fully-populated
columns for indicating finish of code. The single dot or double row of
successive dots 28 in FIG. 2B, 38 in FIG. 3B, and 48 in FIG. 4B indicates
the end of transmission and is printed prior to the symbol for end of
code. The pattern of columns of dots for start and finish of the symbol
may be varied to operate with the array or elements of the reader and also
in timing sequence to allow for precise reading of the symbol.
In the matter of error detection, it can be seen from FIG. 4A, for example,
that the reader 42 is sufficiently wide to cover any combination of eleven
dots and/or spaces to always see a full 9-bit code. In the case of the
letter "C," the reader 42 is moved toward the right and detects a code
pattern for such letter which includes a single dot, a space, six dots and
a space, which includes a space, six dots, a space and a single dot, and
which includes six dots, a space, a single dot and a space. The several
9-bit groups are all detected and are complete patterns for the letter
"C." If a different result is indicated for any one of these patterns, an
error has occurred which may be caused by dirt, a missing dot, or an
electronic failure of a sort. In any event, the operator would be alerted
to sweep the wand across the code a second time. In this manner the error
detection scheme works better when the reader sees at least one full n-bit
code and preferably more than the minimum number of bits required to
correctly identify a perfectly printed pattern. Various methods exist
which use the error detection method as a base to perform error
correction.
It is thus seen that herein shown and described is a high-density dot
matrix code which has both repeated and non-repeated patterns in one
direction representing characters to be read in a direction generally
normal to the patterns. The code and the reading thereof enables the
accomplishment of the objects and advantages mentioned above, and while a
preferred embodiment of the invention (repeated pattern) and a
modification thereof (non-repeated pattern) has been disclosed herein,
other variations beyond those herein mentioned may occur to those skilled
in the art. It is contemplated that all such variations not departing from
the spirit and scope of the invention hereof are to be construed in
accordance with the following claims.
* * * * *
|
|
 |
|
 |
Description |
|
