 |
Description  |
|
|
FIELD OF INVENTION
This invention relates to bar code readers and more particularly to reading
bar codes with a single columnar array.
PRIOR ART
A number of techniques have been utilized reading the various bar code
formats such as UPC, codes having 5 or 7 bars and other codes with various
spacing therebetween.
In one method, width information is derived from a single photo-cell. In
this technique light reflected from a small area of the bar code pattern
is registered by a photo-cell. The analog output of the photocell is
thresholded so that pulse width is a representation of the bar width. The
pulse widths are then measured by various techniques to establish a
relative measure of bar width. The method is sensitive to short term speed
variations and must be compensated for in the recognition process.
In another process the width is derived from an area array. A segment of a
bar code pattern is imaged upon a self-scanned array. Threshold logic is
used to convert the image to a black and white pattern. The bar widths
must be derived from the pattern. In this system, there is a relatively
low frame rate so only a few evaluations of bar widths may be made.
Therefore, the error in this measurement technique may be excessive. OCR
video is peaked for a narrower feature or stroke width than within wider
bar code elements thereby introducing possible other errors.
In a third technique a stationary linear array is used. The measurement
technique uses a stationary array upon which an entire bar code is imaged.
The resolution element size must be less than or equal to 1/2 allowable
bar width error. A cell of the array is examined and declared either to be
black or white. The number of contiguous cells of the same color are used
as the measure of bar width. In this technique there is a trade-off
between the number of cells and resolution since the resolution determines
the size of the pattern that may be read.
SUMMARY OF THE INVENTION
This invention relates to a device for reading characters and bar codes
with the same columnar data lift. The characters are read by scanning with
the array perpendicular to the direction of scan. Bars are read with the
array parallel to the direction of the scan. Skew of up to 45.degree. with
the centerline of the barcode (scan direction) is tolerated as long as
both end cells of the array pass within the area of the bar code to be
read.
This process does not require the storage of video information, therefore,
it is implemented as a high throughput finite state machine. The system
generally has imaging optics and a self scanned array. At the output of
the scan array the output is divided into two parallel outputs. One sends
video information to an optical character recognition unit. The other
parallel path goes to a preprocessor for the bar code reader and then to
the bar reader. In general, the bar code reader and the recognition system
may be as that described and claimed in co-pending U.S. patent application
Ser. No. 252,555 filed Apr. 9, 1981. The optical character reader may be
the same as that described in U.S. Pat. No. 4,075,605.
THE DRAWINGS
The features and technical advance represented by the invention will be
better understood after reading the following description of the invention
along with the drawings which are as follows:
FIG. 1 is a block diagram of the system;
FIG. 2 illustrates the scanning of bar codes;
FIG. 3 illustrates the scanning of alphanumeric data;
FIG. 4 illustrates the data structure for a memory used in the bar width
processor;
FIGS. 5a and 5b are block diagrams of the bar width processor;
FIG. 6 is a flow diagram of the bar width processor;
FIG. 7 is a block diagram of the direction finder, and
FIG. 8 is a flow diagram of the direction finder reset of the bar width
processor.
FIG. 9 is a flow diagram of the direction finder.
PREFERRED EMBODIMENT OF THE INVENTION
Referring to FIG. 1 there is illustrated a block representation of the
invention. Bars or alphanumeric data (A) is scanned through an optics
system (B), by a single columnar array (C). The output of the array is fed
both to an optical character recognition unit (D) and to the bar code
recognition system (E), which includes a bar code preprocessor and a bar
code recognition unit. The output of each of the recognition units if fed
to a decision unit (F) which outputs only one of the recognition units
outputting, that which is scanned and recognized as either bar code or
alphanumeric data.
FIG. 2 illustrates the scanning of bar codes by a single columnar array.
The scan direction is made so that the scanning elements sequentially
intercept the bar codes. Scanning may be from either direction.
FIG. 3 illustrates the scanning of alphanumeric data. The scanning is
different from that of reading a bar code wherein the photo sensitive
elements of a scanning array are scanned parallel across the alphanumeric
data.
FIG. 4 is a block representation of the bar width processor memory which is
recursively updated. The memory size is such that all bars within a scan
frame may be represented in the memory. A scan frame represents a
sequential scanning of the linear array. As each element is scanned again,
a second frame is generated. As illustrated in FIG. 4 "BARS" represents a
pointer to the bar nearest cell 1 of the array. Depending upon scan
direction, this is either a bar about to exit the scan window or a bar
which has just entered, since cell 1 is the first element of the array
clocked out. Type is 1 or 0 and is used as the sign of the result of
count/passes equals width of the bar in cells, which yields an accuracy
better than the plus or minus 2 cell error possible on a given try.
Since the bar has been measured many times, through rapid successive
scanning of the array, and the errors in measurement are random, it is
assumed that the estimated bar width approaches a true bar width, and the
difference between the true bar width and the estimated bar width is less
or equal to the standard deviation for the measurement over N. For
example, error is less than or equal to one cell width divided by N. Given
actual bar width is greater than or equal to one resolution element.
A bar processor has been designed which adds the currently measured bar
widths and increments the number of times the bar has been measured by
cell 1. The tracking of the individual bar through the scan window may be
accomplished using the following.
BARS is always the address of the bar nearest cell 1 of the array. BARS is
incremented or decremented as a bar enters or leaves the end of the scan
window at cell 1. This is indicated by a transition in the color of cell 1
between successive frames. The direction of travel indicates whether to
increment or decrement BARS.
J is the number of transitions (bars) between the bar which is being
measured and the bar nearest cell 1.
BARS plus J therefore indicates an address which is always associated with
a particular bar.
There are three attributes associated with each bar.
(1) The number of counts equals the sum of width measurement for a given
bar.
(2) The number of passes equal the number of times the bar width has been
measured,
(3) The bar type (bar or space between bars) depends upon whether the bar
is black or white.
The output of the bar width processor hereinafter described, is the ratio
of the number of counts to the number of passes with sign equal type. This
information is passed to the bar code recognition unit under the following
conditions:
(1) The bar has exited the field of view from cell 1, for example,
direction equals -1 and a sign change has been made on cell 1.
(2) A new bar has entered the field of view. The memory of the processor is
finite (for example, 16 elements) and addressed in a wraparound or
Modulo-M fashion (mode 16 for this case), therefore, after M(16) bars,
information will be lost unless it is transmitted. It is assumed that M is
larger than the largest number of bars that will enter the field of view
so that the information will be complete on a bar before it is transmitted
to the bar code recognition unit.
(3) Under explicit bar code recognition unit control, if no information has
been passed to the bar code recognition unit within a specified interval,
the recognition unit may examine the contents of the width processor
memory. This would be the case when a bar pattern has been scanned with
the end code still in the scan window and the window stationary. Under
those circumstances the bar code recognition unit would interrogate the
bar width memory.
The final stage of the bar width processor is a divider capable of dividing
a 16 bit word by an 8 bit word. This accommodates 256 passes and allows up
to a 256 element bar within the architecture defined. The sign of the
result is given by the TYPE element so that the recognition unit may
distinguish between black and white bars.
Referring now to FIGS. 5a and 5b, the description of the bar code width
processor is as follows.
A serial string of 1's and 0's representing black and white video from a
linear array enters at 1 accompanied by a cell clock 2 which makes a high
to a low transistion coincident with valid black/white data from the
array. A cycle of this processor is defined as from high to low transition
to the next high to low transition. The video data from the I'th cell of
the array is referred to as cell I. All processing of cell I is
accomplished between the I'th and (I+1) transition of the cell clock
counting from the high to the low transition of start of frame 3.
The LATCH 4 serves as a delay so that cell I present at the output of the
latch, may be compared using COMPARATOR 5 with cell (I+1) just entering
the latch. The second LATCH 6 serves as storage so that cell 1's of
successive frames may be compared. LATCH 4 feeds LATCH 6 which is
indirectly triggered by a start frame 3 via the state control LOGIC 9.
When the COMPARATOR 5 senses old cell 1 not equal to cell (1), it indicates
that the "color" of cell 1 has changed. This implies that cell 1 has
entered or left the area of a bar; for example, crossed a boundary. In the
event that the DIRECTION 14 is positive, a bar has just entered the field
of the array via cell 1. If direction is negative, then a bar just left
through cell 1. In either case, information is output to the bar code
decoder. If direction is +, the bar width information pointed to by
(BARS+1) ie COUNTER 8+COUNTER 10 is output. If direction is negative, than
the information indexed by BARS, ie COUNTER 8 is passed on. In both cases,
the value of the memory locations which have been output are set to zero
after information has been read out.
After processing cell 1, the control logic seeks the first edge or
transition in the array field of view by comparing cell I and cell (I+1).
When cell I is not equal to cell (I+1), J, the relative displacement
COUNTER 10, is cleared to zero, and the WIDTH COUNTER 7 is set to 1. Now
the logic increments the width counter on a cell by cell basis until a
transition is encountered, i.e. Cell (I)=Cell (I+1). The width counter is
then added to the memory indexed by the value of BARS+J 13 by "reading"
the memory using the read line 11, and latching the output of memory into
the ADDER 20. The output of the ADDER 20 is then the total number of cells
counted as associated with the bar represented in memory location
(Bars-J). This is stored using the STORE signal 12. Simultaneously the
term PASSES is incremented by using ADDER 21. Also, simultaneously, the
value of cell I is stored as the TYPE 19 (B/W) of the bar.
Within the same cycle then, J is incremented and width set to 1 before
examination of the next cell.
The process repeats until the entire array has been processed, then repeats
beginning with the examination of cell 1. Note that at the most, 1 bar may
be output during a scan.
The process output is in the OUTPUT of DIVIDER 22; which is the estimated
bar width 23 which is number of counts/number of pasess (17 & 18). The
TYPE 19 is used as sign of the result indicating a black or white bar.
The RESET COMMAND 15 from the direction finder has the following effect: If
the code processor has determined a valid bar has been read, (edit phase)
the reset is delayed until this process is complete. This may be the case
when a valid code is read and the data lift is removed from the paper
immediately. If not in "edit" then the memory for COUNTS 17, PASSES 18 and
TYPE 19 are cleared, BARS 8 is cleared and J (10) is cleared until
direction is stable and valid start of FRAME 3 is received.
To give a more complete description of the processor in FIG. 5 a flow
diagram of the processor operation is illustrated in FIG. 6.
FIG. 7 is a block diagram of the finite state direction finder. In the
direction finder the black and white video and cell clock and start of
frame enter the direction finder section of the bar width processor and
RESET 15 and DIRECTION 14 are generated. RESET 15 occurs each time a
constant change in directions occurs.
At each start of FRAME 3, if cell I has not changed type, for example, a
bar entered or left via cell 1, the first white to black transistion is
located. The location of this transition is labeled TRANS 30. TRANS is
compared to old TRANS from the previous frame's first white to black
transistion. The sign of this comparison is the tenative direction. If the
magnitude of the result is equal to zero then a comparison is made with
the old DIRECTION 31. If direction is not equal to old DIRECTION then
RESET 15 goes active, resetting the bar width processor.
If the direction is the same, then the process is repeated with a new TRANS
30 pending the start of FRAME 3.
If the old cell 1 is not equal cell 1 then the first white to black
transisition is merely stored. In the direction finding process the
direction finder is an implementation of a simple scheme to determine
whether bars are entering or leaving at cell 1 of the array. The process
is designed such that no change is made unless sufficient information for
a good decision process exists. In general the process is as follows:
The black to white edge nearest cell 1 is located. On the next frame, the
nearest white to black transition is found. If cell 1 is changed or no
such transition can be found, no decision is made. The two transition
points are compared. If the second transition is nearer cell 1, direction
is negative, for example, bars leave at the cell 1 end of the array. If
the second transition is farther from cell 1 than the first, the direction
is declared plus. If the change of direction is sensed, then the process
is reset and width information erased. This process is repeated.
FIG. 8 is a flow diagram of the direction finder reset of the bar width
processor.
FIG. 9 is flow diagram of the direction finder.
While specific examples of the invention have been given modification and
changes will be apparent to those skilled in the art while changes and
modifications may fall within the scope of the invention as defined by the
following claims.
* * * * *
|
|
 |
|
 |
Description |
|
