 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus for image reading or
processing, which is able to discriminate whether an input image is a
predetermined image.
2. Related Background Art
The technical development for hard copying, for example with copying
machines, has been aimed at faithful reproduction of the original image,
and as a result, there is now made available a high-quality copying
machine which can provide a reproduced image excellent in color
reproduction and not distinguishable from the original image.
Particularly, the development of digital full-color copying machine has
improved the quality of copied images to a level not distinguishable from
the original image by human eyes, and has enabled one to obtain such copy
within a short time with a simple operation.
However, such image reproduction in a form very close to the original image
could have rise to significant problems, as it may be abused for forgery
of banknotes or other valuable securities. Existing image processing
apparatus, such as copying machines, are powerless to detect and prevent
such abuse.
The assignee of the present application already has disclosed technologies
for preventing forgery or particular images in the U.S. patent Ser. No.
351,165, filed May 12, 1989 and in the new U.S. patent application based
on the priority Japanese Patent Application No. 63-267198, filed Oct. 25,
1988.
The former discloses a technology for altering the image forming process in
a case where the original image to be copied is identified as a
predetermined particular image, and for altering the process according to
the level of similarity to the particular image.
Also, the latter application discloses a technology for registering an
image of which copying is to be prohibited by reading said image with a
reader, and for renewing such registration.
However, such technologies still have room for improvement in the
realization of the image processing apparatus capable of effectively
preventing the forgery of particular images, particularly in the detection
of particular images and in the image processing control in response to
such detection.
SUMMARY OF THE INVENTION
In consideration of the foregoing, an object of the present invention is to
provide an apparatus for image processing or reading, capable of precisely
identifying predetermined original images, such as banknotes or valuable
securities, of which faithful reproduction is to be prohibited.
The above-mentioned object can be attained, according to a preferred
embodiment of the present invention, by an image processing or reading
apparatus provided with detection means for detecting the positional state
of an original image, and discrimination means for extracting image data
of a particular part of said original image in response to the detection
by said detection means and discriminating the similarity thereof to a
pre-registered pattern, wherein a predetermined original image is
identified on the basis of said similarity. Such structure allows precise
discrimination of a particular image such as of banknote placed in an
arbitrary position, for example, on an original support plate, and allows
one to alter the image forming process for such particular image alone.
A second object of the present invention is to provide predetermined image
detecting means, adapted for use in an apparatus completing an image
reading operation by plural mechanical scanning operations, for example an
electrophotographic color copying machine.
The above-mentioned object can be attained, according to another preferred
embodiment of the present invention, by an image reading apparatus
provided with scanning means for scanning an image plural times, and
discrimination means for discriminating whether image is a predetermined
image stepwise in said scanning operations, utilizing the outputs of
plural scanning operations of said scanning means.
Still another object of the present invention is to provide an image
processing apparatus capable of securely prohibiting the reproduction of a
predetermined image when it is entered.
The above-mentioned object can be attained, according to still another
preferred embodiment of the present invention, by an image processing
apparatus provided with means for discriminating whether the input image
is a predetermined image, and control means for deactivating a part of the
image forming apparatus in case said discrimination means identifies that
said input image is said predetermined image.
Also according to another embodiment, there is disclosed an image
processing apparatus provided with means for discriminating whether the
input image is a predetermined image, and control means for changing the
process for a part of color components constituting said input image, in a
case where said input image is identified as said predetermined image by
said discrimination means.
Still other objects, features and advantages of the present invention will
become fully apparatus from the following description of the preferred
embodiments, which is to be taken in conjunction with the attached
drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a view showing a banknote placed on an original support table;
FIG. 2 is a schematic view of an apparatus constituting a first embodiment
of the present invention;
FIG. 3 is a chart showing the relation between different modes and four
scanning operations;
FIG. 4 is a block diagram of an image scanner;
FIG. 5 is a timing chart thereof;
FIG. 6 is a block diagram of a window comparator shown in FIG. 4;
FIG. 7 is a circuit diagram of a block processing unit shown in FIG. 4;
FIG. 8 is a circuit diagram of a water-mark detection unit shown in FIG. 4;
FIG. 9 is a circuit diagram of an address decoder shown in FIG. 4;
FIG. 10 is a flow chart of the function of the apparatus;
FIG. 11 is a view showing the principle of positional detection;
FIGS. 12, 13, 14A, 14B, 14C, 15A and 15B are views showing the principle of
pattern matching;
FIG. 16 is a block diagram of a second embodiment;
FIG. 17 is a view showing the basic structure of a digital full-color
copying machine;
FIG. 18 is a block diagram of a third embodiment;
FIG. 19 is a block diagram of a fourth embodiment;
FIG. 20 is a schematic view of said fourth embodiment;
FIG. 21 is a block diagram of judgment means 3114;
FIG. 22 is a block diagram of an address decoder 3115;
FIG. 23 is a timing chart for counters 3401, 3402;
FIG. 24 is a chart showing an example of output of the address decoder
4115;
FIG. 25 is a view showing an example of input and output images;
FIG. 26 is a block diagram of a selector 4110;
FIG. 27 is a timing chart relating to a counter 4402;
FIG. 28 is a flow chart of the fourth embodiment;
FIG. 29 is a block diagram of a fifth embodiment;
FIG. 30 is a view showing the basic structure of a digital full-color
copying machine;
FIGS. 31, 32A, 32B and 33 are flow charts of the control sequence of the
fifth embodiment;
FIG. 34 is a block diagram of a sixth embodiment;
FIG. 35 is a flow chart of the control sequence of the sixth embodiment;
and
FIG. 36 is a block diagram of a seventh embodiment.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Now the present invention will be clarified in greater detail by preferred
embodiments thereof shown in the attached drawings.
In the following embodiments there are shown applications of the present
invention to a copying machine, but the present invention is not limited
to such embodiments and is naturally applicable to various other
apparatus.
1ST EMBODIMENT
The present embodiment is a digital color copying machine for effecting
color recording in plane-sequential manner, and the detection of
predetermined image is conducted stepwise in sequential manner by plural
scanning operations in relation to said plane-sequential recording.
FIG. 2 is an external view of an apparatus constituting the first
embodiment of the present invention.
The apparatus shown in FIG. 2 is composed of an image scanner unit 201 for
reading an original image and applying digital signal processing thereto,
and a printer unit 202 for printing a full-color image corresponding to
the original image read by the image scanner unit 201.
An original document 204, supported between a mirror-faced pressing plate
200 and an original supporting glass plate (hereinafter called platen) 203
is illuminated by a lamp 205, and the reflected light is guided by mirrors
206, 207, 208 and is focused by a lens 209 onto a 3-line sensor
(hereinafter called CCD) 210, which sends the full-color information in
the form of red (R), green (G) and blue (B) components to a signal
processing unit 211. The entire area of the original is scanned by
mechanical movement of the lamp 205 and the mirror 206 with a speed v, and
the mirrors 207, 208 with a speed v/2, in a direction perpendicular to the
electrical scanning direction of the line sensor. The signal processing
unit 211 electrically processes the signals obtained by said scanning
operation to obtain components of magenta (M), cyan (C), yellow (Y) and
black (Bk) which are sent to the printer unit 202. One of said components
M, C, Y, and Bk is supplied to the printer unit 202 at each original
scanning operation in the image scanning unit 201, so that a printout is
completed by four original scanning operations.
The image signal of the component M, C, Y or Bk sent from the image scanner
unit 201 is supplied to a laser driver 212, which in response modulates a
semiconductor laser 213. The emitted laser beam scans a photosensitive
drum 217, by means of a polygon mirror 214, an f.theta. lens 215, and a
mirror 216.
A rotary developing unit 218 is composed of a magenta developing station
219, a cyan developing station 220, a yellow developing station 221 and a
black developing station 222, which are in succession brought into contact
with the photosensitive drum 217, thereby developing electrostatic latent
images formed thereon with toner.
A sheet fed from a sheet cassette 224 or 225 is wound on a transfer drum
223, and receives the transfer of the image developed on the
photosensitive drum 217.
After four color images of M, C, Y and Bk are transferred in succession,
the sheet is discharged through a fixing unit 226.
The present embodiment will be explained in detail with respect to the
prevention of forgery of a Japanese 10,000-Yen banknote (or bill), as an
example of the predetermined image, but the present invention is naturally
not limited to such embodiment, and is applicable to the prevention of
forgery of other predetermined images such as other banknotes, valuable
securities, contract documents and so on.
In the present embodiment, a copying operation is completed by four
scanning operations as explained above, and the function of the image
scanner unit 201 and the printer unit 202 in each scanning operation are
shown in FIG. 3.
In case of prevention of forgery of a 10,000-Yen banknote, in the first
scanning operation, the image scanner is in a mode 1 for detecting the
approximate position of the 10,000-Yen banknote, and the printer unit
forms the magenta image.
In the second scanning operation, the image scanner is in a mode 2 for
detecting the exact position and angle of the banknote.
In the third scanning operation, the image scanner is in a mode 3, for
calculating the position of the red stamp mark of 10,000-Yen banknote from
the position and angle detected in the second scanning operation, and
after scanning extracts the data of said red stamp mark and discriminates
whether it is actually the red stamp mark. At the same time the printer
unit forms a yellow image.
In the fourth scanning operation, the image scanner is in a mode 4, and
effects the measure for prevention of forgery, if the forgery is
identified to be intended in the third scanning operation.
FIG. 5 is a timing chart showing the function of various parts of the image
scanner of the present embodiment.
A VSYNC signal, indicating the effective image section in the sub scanning
direction, assumes a level "1" in the sections for effecting the image
scanning operations for C, M, Y and Bk in succession. An effective image
section signal VE in the main scanning direction assumes a level "1" where
the image is rendered effective. A main scanning synchronization signal
HSYNC regulates, in the section of level "1", the start position of the
main scanning operation. A pixel synchronization (clock) signal CLK causes
the transfer of the image data at the upshift from "0" to "1". A signal
CLK16 regulates the timing of a 16.times.16 block processed signal to be
explained later, at the upshift from "0" to "1".
IMAGE SCANNER UNIT
FIG. 4 is a block diagram of the image scanner unit 201, in which CCD's
210-1, 210-2, 210-3 having spectral sensitivities respectively in red,
green and blue colors generate signals digitized in 8 bits or levels from
0 to 255.
Since said sensors 210-1, 210-2, 210-3 are mutually spaced as shown in FIG.
1, delay elements 401, 402 are provided for compensating the spatial
aberration.
Logarithmic converters 403, 404, 405 are composed of look-up table ROM's
and convert luminance signals into density signals. A known masking and
undercolor removal (UCR) circuit 406 generates, from input signals of
three primary colors, output signals of Y, M, C, and Bk with a
predetermined number of bits (for example, 8 bits), at each scanning
operation.
An OR gate 407 calculates the logical OR (logical sum) of the output of the
circuit 406 and the value stored in a register 408. The register 408
normally has a value OO.sub.H so that the output of the circuit 406 is
transmitted to the printer unit, but, in case of a forgery preventing
operation, a value FF.sub.H is set in the register 408 through a data bus
from a CPU 417, thereby providing an image uniformly filled with toner.
The CPU 417 controls the apparatus in various modes thereof. A window
comparator 408 discriminates the input of a signal of a predetermined
level designated by the CPU 417, thus detecting the background level in
the modes 1 and 2, or the red stamp mark in the mode 3. A block process
circuit 409 effects processing on every 16.times.16 block of the output
signals of the window comparator 408.
A random access memory (RAM) 412 stores data switched by a selector 411, at
addresses selected by a selector 413.
A main scanning counter 419 is reset by the HSYNC signal, then effects a
count-up operation in synchronization with the CLK signal, and generates
13-bit main scanning (X) addresses X12-X0.
A sub scanning address counter 420 is reset in the section "0" of the VSYNC
signal, effects a count-up operation at the timing of the HSYNC signal,
and generates 13-bit sub-scanning (Y) addresses Y12-Y0.
The CPU 417 effects data writing into and reading from the RAM 412 by
controlling selectors 411, 413, 415, and 416 and an address decoder 414,
according to the operating mode. The CPU 417 is provided with a RAM/ROM
418. A watermark detection circuit 410 serves to detect the watermarked
portion of the banknote.
FIG. 6 is a block diagram of the window comparator 408 shown in FIG. 4,
wherein registers 601, 602, 603, 604, 605, and 606 are connected to the
data bus of the CPU 417 and store values designated by the CPU 417.
Each of comparators 607, 608, 609, 610, 611, and 612 releases an output
signal "1" only when input signals A and B satisfy a condition A>B.
An AND gate 613 releases an output signal "1" only when all the comparators
provide output signals "1", and otherwise releases a signal "0". Thus,
when the registers 601, 602, 603, 604, 605, and 606 respectively store
R.sub.H, R.sub.L, G.sub.H, G.sub.L, B.sub.H, B.sub.L, and the window
comparator 408 releases an output signal "1" only when the input signals
R, G, and B satisfy all the relations:
R.sub.L <R<R.sub.H
G.sub.L <G<G.sub.G
and
B.sub.L <B<B.sub.H
and an output signal "0" otherwise.
FIG. 7 is a circuit diagram of the block process circuit 409 shown in FIG.
4.
Fifteen serially connected fifteen D-flip-flops (DFF) 701-715 serve to
delay the input signal in succession, in synchronization with the pixel
clock signal CLK, and are cleared to "0" in the non-image section where
VE="0".
There are also provided a 5-bit up-down counter 738, an OR gate 733 and an
AND gate 740, which function as shown in the following table.
______________________________________
VSYNC VE X.sub.t X.sub.t-15
Counter output
______________________________________
0 X X X 0 (cleared)
X 0 X X 0 (cleared)
1 1 0 0 Retain
1 1 0 1 Count-down
1 1 l 0 Count-up
1 1 1 l Retain
______________________________________
Thus the output of the counter 738 is cleared to "0" in a period where the
signal VSYNC or VE is "0"; retained when X.sub.t =X.sub.t-15 ; counted up
when X.sub.t =1 and X.sub.t-15 =0; or counted down when X.sub.t =0 and
X.sub.t-15 =1.
In this manner there is obtained the total number 2's in sixteen data
X.sub.t -X.sub.t-15.
Said output is further supplied to first-in-first-out (FIFO) memories
721-735 of one line each, whereby the data of 16 lines are simultaneously
supplied to and added in an adder 741. Thus the total number SUM of 1's in
a window of 16.times.16 pixels is obtained, ranging from 0 to 256.
A digital comparator 742 compares the output SUM of the adder 741 with a
reference value TW predetermined by the CPU 714 and releases the result
"1" or "0".
In this manner the noise elimination in a block of 16.times.16 pixels can
be achieved by suitably selecting the reference value TW.
FIG. 8 shows watermark detection circuit 410, which simultaneously
processes 19 lines with 18 FIFO memories 819, . . . , 820, . . . , 821.
Ten D-flip-flops are serially connected to each of said 19 lines, as
indicated by 801, 802, 803, . . . , 804; 805, 806, 807, . . . , 808; . . .
; 809, 810, 811, . . . , 812; . . . ; 813, 814, 815, . . . , 816, and
additional nine D-flip-flops, indicated by . . . , 817, 818 are serially
connected to the D-flip-flop 812. All these DFF's are driven by the clock
signal CLK16. A signal "1" is released through AND gates 823, 824, 825
when the DFF's 804, 808, . . . , 812, 816 (19 vertical blocks) and DFF's
809, 810, 811, 812, . . . , 817, and 818 (19 horizontal blocks) provide
output signals "1", and the position in this state is latched and sent to
the CPU 417.
FIG. 9 is a block diagram of the address decoder 414.
Registers 901-909 connected to the data bus of the CPU are respectively
given desired values by said CPU.
There are provided four circuit blocks 910-913, but the following
explanation will be given on the block 910, since they are mutually
identical in structure.
Each of subtractors 914 and 915 provides A - B based on inputs A and B. The
most significant bit ("MSB") of said output serves as a sign bit, which is
"1" if A - B is negative.
Each of comparators 916 and 917 provides an output "1" if inputs A and B
satisfy a relation A<B, but provides "0" if the MSB of the input A is "1",
regardless of the input B.
Thus, when the register 909 stores a value BXY, AND gates 918, 919, and 920
of the block 910 provide outputs:
X1=X-address-RX1,
Y1=Y-address-RY1,
E1=1,
only if the following conditions:
##EQU1##
are satisfied.
Similarly, in the block 911, there are obtained outputs:
X2=X-address-RX2,
Y2=Y-address-RY2,
E2=1,
only if the following conditions are satisfied:
##EQU2##
Similarly, in the block 912, there are obtained outputs:
X3=X-address-RX3,
Y3=Y-address-RY3,
E3=1,
only if the following conditions:
##EQU3##
are satisfied.
Similarly, in the block 913, there are obtained outputs:
X4=X-address-RX4,
Y4=Y-address-RY4,
E4=1,
only if the following conditions:
##EQU4##
are satisfied.
When the values RX1, RY1, . . . , and RX4, RY4 are so selected that only
one (at the maximum) of the conditions (1), (2), (3) and (4) can be
satisfied at one time, OR gates 921, 922, 923, 924, and 925 provide output
signals in the following manner.
The OR gate 921 releases one of X1, X2, X3 and X4.
The OR gate 924 releases a signal "1" when any of said conditions is
satisfied.
The OR gate 925 releases one of Y1, Y2, Y3 and Y4.
The OR gates 922, 923 provides outputs signals in the following manner:
______________________________________
Condition 922 (A17) 923 (A18)
______________________________________
(1) satisfied 0 0
(2) satisfied 1 0
(3) satisfied 0 l
(4) satisfied l l
none satisfied 0 0
______________________________________
PROCESS FLOW
FIG. 3 shows four modes 1-4 of the image scanner unit and the outputs of
the printer unit corresponding to four scanning operations in the present
embodiment.
FIG. 10 is a flow chart showing the control sequence of the CPU.
In a step 1001, the CPU 417 sets the mode 1 for the first scanning
operation. In said mode 1, the printer unit generates the magenta image,
and the approximate position of the watermarked portion of the banknote is
detected. For this purpose, the CPU sets values for detecting the
background of the banknote in the registers 601-606 of the window
comparator 408 shown in FIG. 6.
In such state, a step 1002 starts the first scanning.
FIG. 1 shows a 10,000-Yen banknote placed on the platen. In the first
scanning in the mode 1, the window comparator provides output signals "1"
in the hatched areas, namely in the watermarked area and the peripheral
area.
Since such signals continue in excess of a predetermined number of pixels
in the vertical and horizontal directions in the watermarked area, the
watermark detection circuit 410 latches an address corresponding to a
position (Xc Yc) and sends said address to the CPU 417.
Thus the CPU 417 can know the approximate value of the center (Xc, Yc) of
the banknote.
Then a step 1003 sets the mode 2, in which the selectors 411, 413, 415, 416
are set at A, and the address decoder 414 is given following values:
##EQU5##
In the second scanning in a step 1004, by selecting T approximately equal
to 128 mm in the RAM 412, there can be obtained a bit map in which the
hatched areas alone are "1" in the broken-lined frame. Said dimension 128
mm is selected because the banknote can be contained in the broken-lined
area regardless of the kind or position thereof.
In a step 1005, the CPU 417 detects the exact position and angle of the
banknote from said bit map. In this operation the CPU sets the selectors
411, 413 respectively at C, B for freely reading the data in the RAM 412.
There is adopted a method of detecting four corners of the banknote, as
shown in FIG. 11. More specifically, with the banknote position as shown
in FIG.11 four corners are detected, based on facts that:
##EQU6##
Although the details will not be explained, the conditions (5) and (6), if
suitably employed, allow detection of the four corners regardless of the
angle of the banknote.
Also, the angle .theta. in FIG. 1 can be obtained from the equation:
##EQU7##
Also, the position of the red stamp mark is determined from the thus
obtained position and angle. Since the banknote can be placed upward,
downward, face up or face down, the red stamp mark appears at one of the
positions (X.sub.s1, Y.sub.s1), (X.sub.s2, Y.sub.s2), (X.sub.s3, Y.sub.s3)
and (X.sub.s4, Y.sub.s4) shown in FIG. 1. These positions can be
determined by simple calculations from (X.sub.0, Y.sub.0) and .theta..
Then a step 1006 sets the mode 3 for the third scanning, in which the
selectors 411, 413, 415 and 416 are respectively set at B, A, B and B.
In the address decoder 414, there are set conditions:
##EQU8##
and BXY is selected at a number of pixels approximately corresponding to a
size of 300 mm which sufficiently cover the red stamp mark shown in FIG.
1.
Also, the registers of the window comparator 408 shown in FIG. 4 are given
such values as to generate a signal "1" only in the red color part of the
red stamp mark.
Then a step 1007 executes the third scanning, and the output of the window
comparator 408 is stored in the RAM 412, in four positions shown in FIG.
1, where the red stamp mark may appear.
Subsequently a step 1008 extracts the red stamp mark by an algorithm to be
explained later, and the mark is identified in a step 1009. In the absence
of possibility of forgery, namely in the absence of the red stamp mark, a
step 1010 executes the fourth scanning, and a step 1012 fixes the toners
of four colors Y, M, C and Bk.
On the other hand, if the step 1009 detects the red stamp mark, indicating
the possibility of forgery, a step 1011 executes a measure for preventing
the forgery. More specifically, a value FF.sub.H is set in the register,
which normally stores OO.sub.H, of the window comparator 408 shown in FIG.
4, whereby said signal FF.sub.H is sent to the printer unit, thus
depositing the black toner over the entire surface and disabling proper
copying.
In the foregoing description it is assumed that each scanning operation is
conducted corresponding to the printing operation of the printer unit, but
it is also possible to detect the position of the watermark in a
pre-scanning operation if such prescanning is conducted.
PATTERN MATCHING
In the following there will be given a detailed explanation of the pattern
matching conducted in the step 1008.
Since a banknote has two red stamp marks respectively on its top and bottom
faces, two mark patterns are registered in advance for identifying a
banknote.
When the data of a predetermined area of the original image are stored in
the RAM 412, the CPU 417 executes a pattern matching operation with
reference to the content of said RAM 412, according to a flow chart shown
in FIG. 12. Now the RAM 412 stores four sets of binary data, called areas
1-4, as candidates for the predetermined area.
At first a sequence starting from a step 1102 is conducted on the data of
the area 1. The step 1102 executes a window processing for noise
elimination, in a similar manner as the 16.times.16 block process
explained above.
FIG. 13 shows an example of the binary image 1201 of the area 1, wherein
each square represents a pixel which can be white, represented as blank,
or black, represented as hatched. This image is scanned with a window 1202
of 2.times.2 pixels, and a new black pixel is defined if the number of
black pixels in said window is equal to or larger than 2. This scanning
process provides a noise-eliminated pattern 1203 which is vertically and
horizontally reduced to 1/2 in size. The window at the illustrated
position, containing only one black pixel, is replaced by a white pixel in
1204.
Then a step 1104 calculates the position of center of gravity of the
pattern 1203, through the already known method, by projecting the pattern
1203 in the vertical and horizontal directions.
Then the level of similarity is calculated by standard pattern matching. At
first, in a step 1105, a standard pattern registered in advance as a
dictionary is fetched from the ROM 418 to the CPU. Said standard pattern
is the pattern of the red stamp mark of the banknote, but the simple
comparison of the pattern in the step 1103 with said standard pattern does
not necessarily provide a satisfactory result, since the detected pattern
may be rotated by the angle of the banknote on the platen.
The above-explained situation is shown in FIG. 14. For this reason there
may be stored, as the standard pattern, plural patterns obtained by
rotating the pattern of the red stamp mark at a pitch of several degrees,
and a suitable one selected from said plural patterns. For example, there
can be employed 24 patterns obtained by rotating the red stamp mark in a
range of 0.degree.-360.degree. at a pitch of 15.degree., and the
calculation of similarity can be conducted with sufficient precision, by
selecting one of said plural patterns according to the already detected
angle .theta. of the banknote; namely a pattern of rotation angle of
0.degree. for a range -8.degree..ltoreq..theta.<8.degree., a pattern of
rotation angle of 15.degree. for a range
8.degree..ltoreq..theta.<24.degree., a pattern of rotation angle of
30.degree. for a range 24.degree..ltoreq..theta.<40.degree. and so on.
Since the rotation angle .theta. of the banknote may be different from the
rotation angle of the red stamp mark by 180.degree., it is necessary to
select the standard pattern of 180.degree. when that of 0.degree. is
selected. Likewise it is necessary to select the rotation angle
195.degree. at selecting 15.degree., and the rotation angle 210.degree. at
selecting 30.degree.. Thus the calculation of similarity is always
conducted twice.
Then a step 1106 calculates the similarity or correlation. This can be
achieved using various methods, one of which will be explained in the
following. The extracted pattern (a) is represented by B(i, j) while the
selected standard pattern (b) of a certain rotation angle is represented
by P(i, j), wherein said B(i, j) or P(i, j) assumes a value "1" for a
black pixel or "0" for a white pixel. Using the center of gravity
(i.sub.BC, j.sub.BC) of the pattern B(i, j) and that (i.sub.PC, j.sub.PC)
of the pattern P(i, j), determined in the step 2104 shown in FIG. 12, the
correlation of two patterns is determined by the following equation:
##EQU9##
wherein .sym. means the exclusive sum of P and B. Thus the equation (1)
gives the humming distance of the patterns B(i, j) and P(i, j) when the
centers of gravity thereof are matched. The similarity of both patterns is
larger as COR increases.
In the present embodiment, in order to improve the reliability of
similarity and to minimize the mistaken recognition, the correlation is
obtained by the following equation (2) modified from (1):
##EQU10##
wherein "." indicates a logical product, (ANDing) and P(i-i.sub.PC,
j-j.sub.PC) indicates the di | | |
