JP3506822B2 - Author identification method and device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a writer identifying method and an apparatus for identifying a person using handwriting information when writing a writing pattern.

[0002]

2. Description of the Related Art First, a method of initial registration of a standard pattern in this type of conventional writer identification will be described. Figure 2
0 is IEICE research report PRU 89-118 "Online writer verification using fuzzy template" (19
February 28, 1990) p. 11 is a flowchart showing an outline of a conventional standard pattern creation method used in the writer identification shown in “3. Template Creation Method”.

The operation of newly registering a standard pattern in conventional writer identification will be described below with reference to FIG. First, the writer is asked to write the self-written signature three to five times to obtain a sample time series of measurement data representing the pen movement (step ST630). Next, the number of sample points of these 3 to 5 time series data is made constant, and normalization is performed according to the size of the signature (step ST64).
0). Next, one of these 3 to 5 time series data
Dynamic programming matching (hereinafter, referred to as DP matching) with respect to other time-series data with reference to one of them, and obtains a correspondence relationship with each sample point of the time-series data as a reference (step ST650). .
Next, from the sample point of the other time-series data corresponding to the i-th sample point of the time-series data used as the reference,
The maximum value, the minimum value, the center value ai, and the width bi of the maximum value and the minimum value of the data at the th sample point are obtained (step ST660). Finally, as a fuzzification of the interval [ai, bi], the membership function parameters ci, d
i is determined as follows (step ST670).

[0004] ci = ai di = h · bi (h is an arbitrary coefficient)

The correspondence between the sample points of the two time series data Ps and Pt by the DP matching method is obtained as follows. Now, the sampling point time series data of the two time series data Ps and Pt are respectively Ps = s
1, s2, s3, ..., sI, Pt = t1, t2, t
3, ..., tJ, the dissimilarity d _{st, which} is an index indicating the certainty of the correspondence relationship between these sample points, is determined by the i-th sample point si (1 ≦ i) of the time-series data Ps. ≤
I) and the i-th sample point tj of the time series data Pt
The distance d (i, j) of (1 ≦ j ≦ J) is d (i, j) = ‖
If si-tj /, then g is calculated by the following recurrence formula.
(I, J) is obtained, and both time-series data are associated so that g (I, J) becomes the smallest.

[0006]

[Equation 1]

Finally, a dissimilarity d indicating the certainty of the correspondence between the time series data Ps and the time series data Pt.
_st can be _calculated by the following formula. The dissimilarity d _st is an amount effective as a scale for measuring the degree of matching of the two time-series data was performed DP matching method, as the two time series data are similar to each other dissimilarity d _st values Becomes smaller.

[0008]

[Equation 2]

Next, a method of updating the registered standard pattern in the conventional writer identification will be described. FIG. 21 is a p. 1 diagram of "Online scribble character recognition device considering local glyph variation" disclosed in Japanese Patent Laid-Open No. 6-215191. 7 is a flowchart showing an outline of the method described in “B. Update of personalized dictionary” of No. 7.

The operation will be described below with reference to FIG. First, a writer inputs a character pattern. It is assumed that this character pattern has been input (T-1) times in the past and this time is the Tth input (step ST680). To update the standard pattern stored in the personal dictionary, the function F (T) relating to the number of times of input T with respect to the character pattern is used as a weight between the character pattern and the standard pattern, as shown in the following formula. The assigned weighted average is executed to newly obtain the standard pattern X _d (T) (step ST690), and the standard pattern X _d (T) is registered in the personal dictionary (step ST700).

[0011]

[Equation 3]

In the equation, X _d (T-1) is (T-1)
The time series of the sample point coordinates of the standard pattern updated after the second input is shown, and X _in (T) represents the time series of the sample point coordinates of the character pattern input at the Tth time. Further, a and b in the formula are arbitrary constants.

Here, the standard pattern updating method described above is not related to the writer identification, but can be applied to the updating of the individual dedicated dictionary for the individual identification from the viewpoint of updating the individual dedicated dictionary.

[0014]

Since the conventional method of initial registration of the standard pattern in the writer identification is configured as described above, when the registration pattern is created from a plurality of signatures entered by the writer, Even if there are abnormal patterns that are far from other signature patterns, there is a problem that it is impossible to remove these abnormal patterns.

Furthermore, if the shape of the signature to be registered is simple, it can be easily imitated by a third party. Therefore, the signature to be registered has already been registered. When the signature is similar to the signature of the person, there is a problem in that it is difficult to identify the two and the security is lowered.

Further, since the updating of the standard pattern that can be used even in the conventional writer identification is configured as described above, there is a relatively large difference between the already registered standard pattern and the newly input pattern. If there is, the pattern obtained by the weighted average becomes a pattern that does not reflect individuality at all, and not only the accuracy is lowered, but also it is updated every time the pattern is input, so the processing is useless. There was a problem. Further, since there is no means for motivating the update of the standard pattern, the standard pattern can be freely updated by anyone other than the writer himself, and there is also a problem that the security of the standard pattern of each individual against a third party is poor.

The present invention has been made in order to solve the above problems, and creates a standard pattern of a writer himself / herself to be registered in a writer identification dictionary with high precision and security, and to create the standard pattern of the writer. An object of the present invention is to provide a writer identification method and an apparatus that can be updated while maintaining accuracy and security while updating.

[0018]

A writer identification method according to the present invention compares an input writing pattern with a standard pattern registered in a writer identification dictionary to determine the authenticity of the writing pattern, and When it is determined that the writing pattern belongs to the writer himself, the step of creating a standard pattern based on the information related to the writing pattern is compared with the created standard pattern and the contents of the writer identification dictionary, and the standard and determining the validity of the update of the author identification dictionary by pattern, if it is determined to be valid and a step of updating the author identification dictionary by the standard pattern, the step of creating a standard pattern, the author identification
It is provided with a step of determining whether or not to give permission to create a standard pattern to a writing pattern desired to be registered in the writer identification dictionary from among the writing patterns input when a new standard pattern is registered in the dictionary. It is a thing.

The writer identification device according to the present invention is a writer identification dictionary that registers a standard pattern of handwriting, an input unit that allows a writer to input a writing pattern, a writing pattern of a writer, and a standard registered in the writer identification dictionary. A writer identification means for comparing the pattern with the authenticity of the writing pattern,
The standard pattern creating means for creating a standard pattern from the writing pattern and the registration or updating of the standard pattern created by the standard pattern creating means based on the matching result of the standard pattern created by the standard pattern creating means and the contents of the writer identification dictionary Equipped with standard pattern registration / update determination means,
Permission to create a standard pattern for a writing pattern that the writer desires to register in the writer identification dictionary from among the writing patterns input by the writer when inputting a new standard pattern to the writer identification dictionary It is provided with a writing pattern verification means for judging whether or not to give.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below. Embodiment 1. 1 is a block diagram showing a configuration of a writer identification device according to a first embodiment of the present invention. In the figure, reference numeral 1 is an input means for supporting the input of a writing pattern performed by a writer and the input of other arbitrary data. The output of the input means 1 is a data buffer 2 and a writer identification means which will be described later. 7 and standard pattern update promoting means 11. Further, 8 is a display means for displaying the output contents of each connected means, including the identification result of the writer identifying means 7, when the output thereof is required to be displayed. The input unit 1 and the display unit 8 are composed of, for example, an electronic pen and a display-integrated tablet.

Reference numeral 2 denotes a data buffer, which is used to store the writing patterns entered by using the input means 1, and the pattern selecting means 3 which will be described later on the writing patterns.
Output to. A pattern selecting means 3 selects a writing pattern effective for creating a standard pattern among a plurality of writing patterns stored in the data buffer 2 and outputs it to a standard pattern creating means 4 described later.

Reference numeral 4 is a standard pattern creating means for creating a standard pattern based on the writing pattern selected by the pattern selecting means 3. For example, the number of sample points of each writing pattern selected by the pattern selecting means 3 is set to a predetermined number of sample points N.
If the number of sample points in the writing pattern is less than Ns, the sample points are inserted by linear interpolation or the like to obtain the same number of sample points Ns, and the coordinate value of each sample point is A standard pattern is created by calculating the average.

Reference numeral 5 is a standard pattern registration / update determination means, which collates with the standard pattern already registered in the writer identification dictionary 6 before registering the standard pattern in the writer identification dictionary 6 to be described later, and collates the standard pattern. From the result, it is determined whether or not to register or update, and only when it is determined to register or update, the registration or updating process is performed on the writer identification dictionary 6. Reference numeral 6 is the writer identification dictionary in which individual standard patterns are registered for a plurality of writers.

Reference numeral 7 denotes a writer identification means, which is a writer identification dictionary 6
Patterns and input means 1 of each writer registered in
The same DP as the conventional example between the writing patterns input from
The degree of dissimilarity is checked by the matching method, and the standard pattern that is the most similar, that is, the standard pattern having the smallest degree of dissimilarity, is retrieved from the writer identification dictionary 6, and the input means is determined by the degree of dissimilarity of the standard pattern with respect to a predetermined criterion level. The authenticity of the writing pattern input from 1 is determined.

Reference numeral 9 denotes a control means, which performs various kinds of operation control for giving an operation command in each of the above means, has a timer function, and provides time information to each means as necessary. Reference numeral 10 denotes a standard pattern registration switch, which is turned on when a new standard pattern is newly registered in the writer identification dictionary 6, and is normally turned off at other times. Reference numeral 11 is a standard pattern update accelerating means, which operates when the standard pattern registration switch 10 is OFF, and is registered in the writer identification dictionary 6 based on the identification result returned from the writer identifying means 7. It encourages the writer to update the standard pattern.

The lines with arrows in the figure show the flow of data, and the lines without arrows show the control lines showing the connection with the control means 9.

[0027] In the following, it will be described separately in the operation as "initial registration at the time of the operation of the standard pattern" and "normal operation at the time."

[0028] 1. Operation during Initial Registration of Standard Pattern Here, FIG. 2 is a flowchart showing a flow of processing when initial registration of a standard pattern is performed in the writer identification dictionary 6 in the writer identification device shown in FIG. First, the writer turns on the standard pattern registration switch 10. This allows the writer to register the standard pattern for the first time (step ST10). Next, the writer inputs the writing pattern from the input means 1 by writing the writing pattern on the tablet using the electronic pen (step ST20). The input unit 1 detects the coordinate points of the locus drawn on the tablet for each predetermined sampling period while writing the writing pattern on the tablet.
These coordinate values are output to the data buffer 2 and stored (step ST30). The writing pattern stored in the data buffer 2 is sent to the pattern selecting means 3, where the writing pattern effective for creating the standard pattern is selected (step ST40). Next, it is determined whether or not a predetermined number of writing patterns have been collected (step ST50), and if not, step ST2.
After returning to 0, the writing pattern is input again.

On the other hand, if the gathered writing patterns required number are standard pattern generating means 4 carries out the creation of the standard patterns from the writing pattern selected (step S
T60). Next, the standard pattern registration / update determination means 5 collates the standard pattern created in step ST60 with the contents of the writer identification dictionary 6 to check, for example, whether it is erroneously identified by another standard pattern already registered. That is, it is judged that there is no adverse effect only when any standard pattern already registered is rejected without being accepted, and when no standard pattern is registered in the writer identification dictionary 6, and The standard pattern is registered in the writer identification dictionary 6 (step ST70). Finally, the standard pattern registration switch 10 is turned off and the standard pattern registration processing is completed (step ST8).
0).

FIG . 3 shows a writing pattern input process, that is, a pattern selection process 100 effective for creating a standard pattern in the pattern selection means 3, which is surrounded by a dotted line shown by reference numeral 100 in FIG. It is a flow chart showing a flow of detailed operation. In FIG. 3, steps ST20 and ST30 are the same as those in FIG.
Since the operation is the same as those of the above, the description thereof will be omitted. Note that, in FIG. 3, the portion surrounded by the dotted line indicated by reference numeral 3 indicates the operation of the pattern selection means 3 in FIG.

The pattern selection processing effective in creating the standard pattern in the pattern selection means 3 will be described below with reference to FIG. In order to select a pattern having the highest number of strokes from the writing patterns input through the processing of steps ST20 and ST30, the reference numeral 20
The processing of the part surrounded by the one-dot chain line indicated by 0 is executed. That is, first, the input writing pattern is classified according to the number of strokes (step ST90). Next, among the writing pattern groups classified by the number of strokes, the writing pattern group of the number of strokes with the highest frequency of appearance is selected. At this time, it is assumed that the number of strokes of the writing pattern group is k1 and that the number of strokes is n1 (step ST100).

[0032] Then in order to test the amount of information possessed by the writing pattern is compared with the required minimum number of strokes KAKU the stroke count k1 of the selected writing pattern group (step ST110). At that time, if the stroke number k1 of the selected writing pattern group does not reach the required minimum stroke number KAKU, it is determined that the features necessary for identification cannot be sufficiently extracted, and all the input writing patterns are removed (step ST170). Step ST
Return to 20 and try again. On the other hand, if the stroke number k1 of the selected writing pattern group is equal to or larger than the required minimum stroke number KAKU, the total number n1 of patterns of the selected writing pattern group,
The number of patterns N1 required for creating a predetermined standard pattern is compared (step ST120). As a result, when the total number of patterns n1 is less than the required number of patterns N1, the process returns to step ST20 and additional writing patterns are input. The above operation is performed until the total number of patterns n1 becomes equal to the required number of patterns N1.

The total number of patterns n1 is the required number of patterns N1
Then, the abnormal pattern that is not effective for creating the standard pattern is removed. The part surrounded by the alternate long and short dash line indicated by reference numeral 300 in FIG. 3 corresponds to the processing. Here, first, regarding the total stroke length of each writing pattern, the average value A for all N1 writing patterns and the i-th writing pattern of all N1 writing patterns were excluded (N1-1). For each writing pattern, a change amount Di (1 ≦ i ≦ N1) defined by the following equation is obtained from the average value Ai (1 ≦ i ≦ N1) (step ST130).

[0034]

[Equation 4]

[0035] Then each writing pattern variation found for Di (1 ≦ i ≦ N1) , and compares the threshold value α that has been determined in advance. At this time, if the condition of “Di> α” is satisfied, the i-th writing pattern is removed from the selection as an abnormal pattern and removed (step ST140).
The reason why a writing pattern satisfying the condition of “Di> α” can be determined as an abnormal pattern is that if all writing patterns have similar characteristics, the average value of all writing patterns and the i-th There should be almost no difference between the two in the average value obtained by excluding the writing pattern.
On the other hand, when a large difference occurs between the two average values, the deviation of the i-th writing pattern from other writing patterns is large. Therefore, at this time, the i-th writing pattern is determined to be an abnormal pattern. It becomes possible to do.

[0036] At this time, to be carried in the removal of abnormal patterns, since writing pattern has been removed does not exist, writing pattern number n1 that is selected, but remains kept at required pattern number N1 to create standard patterns If the abnormal pattern is removed, the number of selected patterns n1 becomes less than the required number of patterns N1. In such a case, the process returns to step ST20 again, and the writing pattern is additionally input (step ST150).

On the other hand, if the removal of abnormal pattern is not performed in step ST150, i.e. variation Di
When there is no writing pattern in which (1 ≦ i ≦ N1) exceeds the threshold value α, as a next step, determination based on the shape of the writing pattern described later is performed to prevent registration of a simple shape pattern (step ST160). . If it is determined in step ST160 that the shape is simple, all the input writing patterns are erased (step S
(T170), the process returns to step ST20 again, and the writing pattern is input again from the beginning.

FIG . 4 is a flow chart showing a detailed operation flow for shape determination by the processing for preventing the registration of the writing pattern having a simple shape in step ST160 of FIG. Further, FIG. 5 is a flowchart showing a detailed operation flow of processing for calculating the curvature of each stroke forming the writing pattern in FIG. The operation of preventing the registration of the writing pattern having a simple shape will be described below with reference to FIGS. 4 and 5.

First, referring to FIG . 4, step ST1 of FIG.
In the change amount Di (1 ≦ i ≦ N1) obtained in 30,
The i-th writing pattern showing the minimum value is selected (step ST180). Thereafter, the writing pattern selected in step ST180 is processed.
Next, for each stroke that constitutes the writing pattern selected in step ST180, the curve degree Cj (1 ≦ j ≦ k1) is obtained based on the processing described later shown in FIG. 5 (step ST190). Next, the curve degree Cj obtained for each stroke is set to a predetermined threshold value T
Compared with H_CURVE, the curve degree Cj is the threshold value TH_C.
The sum of stroke lengths that are less than URVE is calculated, and it is determined whether or not it exceeds 1 / n of the stroke length of the entire writing pattern (step ST200). The above n is a predetermined integer.

In this step ST200, the length of the stroke at which the curve degree Cj becomes less than the threshold value TH_CURVE is
When it is determined that the stroke length exceeds 1 / n of the stroke length of the entire writing pattern, it is determined that the writing pattern has a simple shape (step ST210), and step ST in FIG.
The control is passed to 170, and this processing ends. If the length of the stroke at which the curve Cj is less than the threshold value TH_CURVE is 1 / n or less of the stroke length of the entire writing pattern, it is determined that the pattern is not a simple shape, and the process of the pattern selecting unit 3 is terminated. Yes (step ST22
0).

[0041] Hereinafter, with reference to the flowchart of FIG. 5 describes the processing for calculating the curve of Cj of each stroke. First, the counter variable j is initialized to 1 (step ST230). Next, the counter variable j is compared with the input stroke number k1, and if the counter variable j is less than or equal to the input stroke number k1, the processing is continued, and if the counter variable j exceeds the input stroke number k1, this processing is ended (step ST240). .

If the counter variable j is less than or equal to the number of input strokes k1, first the length Lj and the straight line distance Mj between the start and end points of the jth stroke of the writing pattern are obtained (step ST250), and then both of them. Ratio of values Lj
The curve degree Cj is calculated by / Mj (step ST26).
0). Finally, the counter variable j is incremented and the process returns to step ST240 (step ST270), and this process is repeated until the counter variable j exceeds the input stroke number k1.

Next, in step ST70 shown in FIG. 2, the operation of the standard pattern registered / update deciding section 5 for registration determination reference pattern will be described. Figure 6 is a <br/> flowchart shows the flow of operation of the standard pattern registration / update deciding section 5, strictly speaking, numeral 5 in the figure
The part surrounded by the dotted line indicated by the symbol corresponds to the operation of the standard pattern registration / update determination means 5.

[0044] Hereinafter, the operation of the standard pattern registered / update deciding section 5 with reference to FIG. First, the standard pattern created by the standard pattern creating means 4 is compared with another standard pattern already registered in the writer identification dictionary 6 (step ST280). The DP matching described in the conventional example is used for this matching method. That is, the standard pattern showing the minimum dissimilarity is selected from the already registered standard patterns in the writer identification dictionary 6, and if the minimum dissimilarity is equal to or less than a predetermined threshold value, the standard pattern creating means 4 It is determined that the created standard pattern is accepted with respect to the standard pattern already registered in the writer identification dictionary 6 showing the minimum difference. On the other hand, if the minimum dissimilarity exceeds the threshold value, it is determined to reject. Here, if there is no standard pattern already registered in the writer identification dictionary 6, there is no standard pattern that can be matched, and therefore a rejection is determined.

The next standard pattern registration switch 10 is ON
Is checked (step ST290), O
In the case of N, it branches to step ST300, and in the case of OFF, it branches to step ST330. Since the standard pattern registration switch 10 is ON at the time of initial setting of the standard pattern, the process proceeds to step ST300. For the operation when the standard pattern registration switch 10 is OFF, refer to “2. Normal operation” later.
It will be described when explaining.

In step ST300, step ST2
Based on the standard pattern matching result obtained in 80, it is determined whether or not any standard pattern registered in the writer identification dictionary 6 has not been accepted, that is, whether or not the input has been rejected. As a result, only when the rejected matching result is obtained, the standard pattern created by the standard pattern creating means 4 is registered in the writer identification dictionary 6, and the registration date and time is also recorded (step ST310). On the other hand, if any of the standard patterns already registered in the writer identification dictionary 6 is accepted, it is unsuitable for newly registering because a similar standard pattern already exists,
A sentence indicating that it has not been registered is output to the display means 8 and the whole process is terminated (step ST320).

[0047] 2. Operation in Normal Time FIG. 7 is a flowchart showing a flow of operation in normal time when the standard pattern registration switch 10 in the writer identifying apparatus shown in FIG. 1 is turned off, and FIG.
Explanatory diagram showing a display example of the message displayed by
FIG. 9 is a flowchart showing the flow of the operation of the writer identifying means 7, and FIG. 10 is an explanatory view showing the contents of the writer identifying dictionary 6. Note that, in FIG. 7, a portion surrounded by a dotted line indicated by reference numeral 11 corresponds to the operation of the standard pattern update promoting means 11. Further, in FIG. 7, steps ST20 to ST6
Since 0 is the same operation as those shown in FIG. 2, description thereof will be omitted.

The normal operation will be described below with reference to FIGS. 7 to 10. In FIG. 7, the writer first inputs the writing pattern using the input means 1 (step ST360). The writing pattern input to the input unit 1 is input to the writer identifying unit 7, and the writer is identified with respect to the writing pattern (step ST370). As a result, if the writing pattern is accepted by the writer identifying unit 7, the identification result is output to the standard pattern updating promoting unit 11, and if rejected, the process ends (step ST380).

In the operation of the standard pattern update promoting means 11 surrounded by the dotted line indicated by reference numeral 11 in FIG . 7, the degree of difference when the writing pattern is accepted from the identification result output by the writer identifying means 7. D is compared with a predetermined threshold value TH_DIS_LOW, and the difference D is the threshold value TH.
If it exceeds _DIS_LOW, it is determined that the writing pattern is largely deformed from the standard pattern registered in the writer identification dictionary 6, and the process proceeds to step ST400. On the other hand, if the dissimilarity D is less than or equal to the threshold value TH_DIS_LOW, the writing pattern is still the author identification dictionary 6
It is determined that the deviation from the standard pattern registered in is not so large, and the process is terminated without prompting the standard pattern update (step ST390). The difference D when the writing pattern is accepted is the threshold TH_D.
If IS_LOW is exceeded, a message prompting to update the standard pattern registered in the writer identification dictionary 6 is output to the display means 8 (step ST400), and the standard pattern registered in the writer identification dictionary 6 is updated. Ask the writer whether to implement it (step ST410).

FIG . 8 shows an output example of a message for prompting the updating of the standard pattern output to the display means 8 in step ST400 of FIG. In the figure, 12 is a display screen of a display-integrated tablet, for example, and 13 shows the contents output by the standard pattern update promoting means 11. Reference numeral 14 is a YES button selected by the writer when affirming the content of the document indicated by 13, and 15 is a NO button selected by the writer when rejecting the content of the sentence indicated by 13.
That is, if the writer wants to update the standard pattern registered in the writer identification dictionary 6, for example, the YES button 14 is selected with the electronic pen, and if the update is not desired, the NO button 15 is selected. .

If the writer selects the YES button 14 and wishes to update the standard pattern, the process proceeds to step ST20 and thereafter to create a new standard pattern. On the other hand, if the NO button 15 is selected and the standard pattern is not desired to be updated, the process is terminated.
Note that steps ST20 to ST60 are the same operations as those in FIG. 2 described above, and therefore description thereof will be omitted. When the standard pattern is created in step ST60, the standard pattern registration / update determination means 5 then determines the validity of the update of the created standard pattern (step ST420).

In order to explain the processing of the last step ST420 in FIG. 7, the explanation will be returned to FIG. 6 showing the operation of the standard pattern registration / update determination means 5. The difference from the above description of the operation in “1. Initial registration of standard pattern” is
The standard pattern registration switch 10 is being turned off. Since the standard pattern registration switch 10 is off, the process proceeds from step ST290 to step ST330. In this step ST330, the difference D when the standard pattern newly created by the standard pattern creating means 4 is accepted to be the writer himself and the threshold value TH_DIS_LOW are compared. As a result, if the dissimilarity D is less than or equal to the threshold value TH_DIS_LOW,
Assuming that there is no difference between the newly created standard pattern and the standard pattern of the writer himself / herself already registered in the writer identification dictionary 6, a message to the effect that updating is not necessary is output to the display means 8 and the process is terminated (step ST350). When the dissimilarity D exceeds the threshold TH_DIS_LOW or is rejected (when the dissimilarity D indicates the maximum possible value), the standard pattern already registered in the writer identification dictionary 6 is used. Assuming that the difference with
The newly created standard pattern is additionally registered and the date and time of the addition is also recorded (step ST340).

Next, with reference to FIGS. 9 and 10, in step ST370 in FIG. 7, the operation of the author identification process will be described when performing the identification of the input handwritten pattern. Note that this corresponds to the operation performed by the writer identifying means 7 and the writer identifying dictionary 6 in the writer identifying device shown in FIG.

[0054] First, in FIG. 9, sample number N which predetermined sample number of the input handwritten pattern
If the number of sample points of the input writing pattern is less than Ns, the number of sample points is thinned out to s, or the sample points are linearly interpolated and inserted to make the number of sample points Ns (step ST430). Next, a standard pattern written with the same number of strokes as the input writing pattern is selected from the writer identification dictionary 6 (step ST440). This step ST440
DP matching based on sample point coordinate values is performed between each standard pattern selected in step 1 and the input writing pattern, and the difference Ds from the input writing pattern is calculated for each selected standard pattern. Calculate (Step S
T450).

[0055] dissimilarity m of standard patterns whose value in the difference degree Ds obtained in step ST450 becomes minimum
in (Ds) is set to a predetermined threshold value TH_DI
If it is compared with S_S (step ST460) and the minimum dissimilarity min (Ds) is less than or equal to the threshold value TH_DIS_S,
As the writer of the standard pattern showing the minimum difference min (Ds), a sentence indicating that the writing pattern has been accepted is output to the display means 8 (step ST470). Then, the date and time when it is accepted is recorded in the writer identification dictionary 6 (step ST480). Also, the minimum dissimilarity min
If (Ds) exceeds the threshold value TH_DIS_S, a sentence indicating that the input writing pattern has been rejected is output to the display means 8 and the process ends (step ST4).
90).

[0056] shows the contents of this author identification dictionary 6 in Figure 10. The writer identification dictionary 6 is configured so that which standard pattern the writing pattern input from the input means 1 is accepted can be managed by a registration number. The number of registrations indicates the number of registered standard patterns, and a plurality of different standard patterns can be registered with one registration number. In addition, the date and time of registration of each standard pattern and the date and time of reception of the writing pattern previously input from the input means 1 using the standard pattern are recorded.
In the figure, the part indicated by 1000 shows one standard pattern, and there are as many standard patterns as there are registration patterns. The registration number is incremented each time a new standard pattern is registered.

[0057] As described above, according to the first embodiment, target
At the time of initial input of the quasi-pattern, the input writing pattern
The standard pattern created by selecting a valid one from the
When it is judged that the registration is valid by comparing with the contents of the identification dictionary
Then, register the standard pattern in the author identification dictionary and
When the code is updated, the input writing pattern is the writer's book.
If it belongs to a person, it is based on the information related to the writing pattern.
Based on this, the standard pattern is prompted to be updated, and an update request is made.
And select a valid one from the input writing patterns
Match the created standard pattern with the contents of the writer identification dictionary.
Configured to determine the relevance of the update, the standard
When updating the pattern, the registered writer himself
A writing pattern that is identified as the same as the standard pattern
Since it is necessary to enter it once, the standard of the writer himself
The security of the pattern is maintained and its identification
Update from the fact that the standard pattern is prompted based on the result
Is not always done and only at the request of the scribe
It will be implemented, and unnecessary processing will not be done
Mu. Also, create a standard pattern by using an effective writing pattern.
Because you choose and are based on that,
You can create a standard pattern with a high degree of accuracy, and
Matches the quasi-pattern with the contents of the author identification dictionary and
Register or update the standard pattern based on the result
Since there is no unnecessary registration or update,
The effect is that the accuracy of the original standard pattern can be maintained.
It

[0058] Also, by selecting the writing pattern of most frequent strokes by the pattern selecting means 3,
This has an effect of facilitating removal of an abnormal pattern, that is, a writing pattern that is not effective for creating a standard pattern.

[0059] In addition, the pattern selection means 3
The brush is based on the average value of the shape features of the pattern.
Is the pattern described above effective for creating a standard pattern?
or not The writing pattern that determines the
To determine that a pattern is not valid for creating a standard pattern
Since it is configured to, abnormalities that are not effective for creating standard patterns
There is an effect that the pattern can be removed accurately.
It

In addition, the pattern selecting means 3 determines, for example, all N for all stroke lengths of each writing pattern.
The average value A of one writing pattern and the average value Ai (1 ≦ i ≦ of the (N1-1) writing patterns of all N1 writing patterns excluding the i-th writing pattern. Change amount Di defined by N1)
By using (1 ≦ i ≦ N1), a pattern that is not effective in creating a standard pattern, that is, an abnormal pattern is detected. Therefore, the judgment is performed from the statistical distribution of the shape features of the writing pattern. There is an effect that an abnormal pattern can be detected well.

[0061] Further, since the selection of the writing pattern evaluates the complexity of the shape of the writing pattern from the curve degree Cj of each stroke in the pattern selecting means 3 and the row Migihitsuji, simple writing pattern shape as a standard pattern It becomes easy to prevent registration in the writer identification dictionary 6,
This has the effect of improving security for third parties.

Further , when the abnormal pattern is eliminated by the pattern selecting means 3, the writing patterns are additionally inputted by at least the excluded pattern so as not to reduce the number of writing patterns necessary for creating the standard pattern. it is possible to create a high standard pattern
There is an effect .

[0063] Further, at the time of initial registration of the standard pattern, on a standard pattern created by the reference pattern generator means 4 was confirmed by standard pattern registration / update determination unit 5 that nothing similar in my identification dictionary 6 in since be registered to the authors identification dictionary 6, there is no already be accidentally registered when the standard pattern similar to the author identification dictionary 6 are registered, the effect of improving the security.

[0064] Also, during normal operation, i.e., the standard when the pattern registration switch 10 is in the OFF, the standard pattern registration / update determination means by the 5 author identification dictionary standard patterns created by the reference pattern generator means 4 6 standard pattern of scribes person that is registered as against the, already registered standard pattern created before <br/> SL also the author identification dictionary 6 in any of the case where it is the case or received is denied Since it is decided whether to update based on the degree of dissimilarity to the standard pattern of the writing writer himself, there is no need to constantly update, unnecessary processing is not done, and it is updated only when necessary. Therefore, there is an effect that accuracy is maintained .

[0065] Also, during normal operation, i.e., when the standard pattern registration switch 10 is in the OFF, the standard pattern updating facilitating means 11 for the first time when the author identification means 7 the writing pattern input by the input means 1 has been received It works and is obtained by the writer identification means 7 at that time
Wherein the dissimilarity to the standard pattern of the writing in person the author identification dictionary 6 of the writing pattern, author identification dictionary 6
Since the update of the standard pattern registered in is promoted, the confidentiality of the standard pattern of the writer himself / herself registered in the writer identification dictionary 6 is improved, high security is maintained, and corner in is not performed unnecessary processing because there is possible to determine the updating of the reference pattern, also has the effect of accuracy is maintained because only update prompt when necessary is carried out.

In the first embodiment, the total stroke length of the writing pattern is used as a feature when the change amount Di is obtained in step ST130 of FIG. 3 at the time of initial registration of the standard pattern, but the writing pattern is formed. The change amount Di may be obtained for each stroke, and the writing pattern having the stroke satisfying the condition of “Di> α” may be excluded between the change amount Di and a predetermined threshold value α.

[0067] Also, as the change amount Di is eliminated and the sum of the length of the satisfying stroke of the "Di>alpha", the writing pattern the ratio of the stroke length of the whole writing pattern becomes more arbitrary constant β You may

Further , when the change amount Di is obtained in step ST130 of FIG. 3, the average value A of all N1 writing patterns and (N1-1) writings excluding the i-th writing pattern. Although the absolute value of the difference between the average values Ai of the patterns is standardized by the average value A of all the N1 writing patterns, the i-th value is obtained from the variance value σ of all the N1 writing patterns. The value obtained by subtracting the variance value σi of the (N1-1) writing patterns excluding the writing pattern is normalized by the variance value σ of all the N1 writing patterns, and the value is positive. When the constant γ of is exceeded,
That is, if the variance when the i-th writing pattern is removed is smaller than that before removal and the reduction amount exceeds γ, the i-th writing pattern may be removed as an abnormal pattern. Good.

Further , in the operation of step ST260 in FIG. 5, the curve degree Cj of each stroke is defined by the ratio of the j-th stroke length and the distance between the start and end points of that stroke. The directions between the points are classified into eight directions, for example, upper, lower, left, right, upper left, lower left, upper right, lower right, and the length of the j-th stroke is sorted for each direction, It may be defined as a ratio of the j-th stroke in the longest direction to the total length.

In the normal operation, in the first embodiment, as a method of updating the standard pattern registered in the writer identification dictionary 6, a new standard pattern is additionally registered. The standard pattern may be deleted and rewritten with a newly created standard pattern.

[0071] Embodiment 2. FIG. 11 is a block diagram showing the configuration of the writer identification apparatus according to the second embodiment of the present invention. As in the case of the first embodiment shown in FIG. 1, 1 is input means, 2 is a data buffer, and 3 is Pattern selecting means, 4 is a standard pattern creating means, 5 is a standard pattern registration / update judging means, 6 is an author identification dictionary, 7 is an author identifying means, 8 is display means, 9 is control means, 10 is a standard pattern registration switch. Shows. Further, reference numeral 16 denotes the first embodiment indicated by reference numeral 11 in FIG.
Is a standard pattern update promoting means corresponding to the standard pattern update promoting means in FIG.
It operates when 0 is OFF, and prompts the writer to update the standard pattern registered in the writer identification dictionary 6 based on the identification result of the writer identification means 7 and the contents of the writer identification dictionary 6. Is.

[0072] FIG. 12 is a normal operation in the author identification apparatus shown in FIG. 11, that is, a flowchart showing a flow of operation when the standard pattern registration switch 10 is OFF. In the figure, the part surrounded by the dotted line indicated by reference numeral 16 corresponds to the operation of the standard pattern update promoting means 16. Note that steps ST360 to ST3
80, steps ST400, ST410, step ST
20 to ST60, and step ST420, FIG.
The same processing as those steps in the first embodiment shown in FIG. Further, the operation at the time of initial registration of the standard pattern in the writer identifying apparatus shown in FIG. 11, that is, the operation when the standard pattern registration switch 10 is ON, is the same as that in the case of the first embodiment shown in FIG. Since they are the same, the description thereof will be omitted.

[0073] Next, the operation will be described. In the operation of steps ST360 and ST370 in FIG.
When the writing pattern input from is accepted by the writer identifying means 7 (step ST380), the recognizing result is input from the writer identifying means 7 to the standard pattern update promoting means 16. Receiving the standard pattern update promoting means 16
Searches the author identification dictionary 6 for the last received date and time of the information of the standard pattern determined to be the same as the writing pattern and the information of the latest date and time of the registered date and time of the standard pattern. (See Figure 10). Then, the writing pattern is compared with the current date and time at which the writing pattern is accepted, and it is determined whether or not a predetermined time TIME or more has elapsed (step ST50).
0).

When it is determined in step ST500 that the predetermined time TIME or more has elapsed, the screen shown in FIG. 8 is displayed on the display means 8 based on the processing in steps ST400 and ST410. Ask the writer whether to update the standard pattern registered in the writer identification dictionary 6. The writer is YE
Only when the S button 14 is operated and the writer identification dictionary 6 is desired to be updated, the processes of steps ST20 to ST60 and step ST420 are executed to update the standard pattern. On the other hand, when the above condition is denied, the processing is ended as it is.

[0075] As described above, according to the second implementation, the standard pattern updating promoting means 16 first operates when the author identification means 7 the writing pattern input by the input means 1 has accepted, also the author of that time Based on the registration date and time of the standard pattern of the writer himself / herself in the writer identification dictionary 6 that the identification means 7 has determined to be the same, or the date and time of last acceptance, the updating of the standard pattern registered in the writer identification dictionary 6 is prompted Therefore, the confidentiality of the standard pattern of the writer himself / herself registered in the writer identification dictionary 6 is improved, the high security is maintained, and the standard pattern is not always requested to be updated. corner without useless processing is performed, the effect of the accuracy of the standard pattern is maintained.

In the second embodiment, the standard pattern registered in the writer identification dictionary 6 is updated by newly registering the standard pattern. However, in the case of the first embodiment. Similarly, the existing standard pattern may be deleted and rewritten with a newly created standard pattern.

Embodiment 3. FIG . 13 is a block diagram showing the configuration of the writer identification apparatus according to Embodiment 3 of the present invention, wherein 1 to 10 in FIG.
Alternatively, they are the same as those in the first or second embodiment shown by attaching the same reference numerals in FIG. Further, 17 is the standard pattern update promoting means 11 in the first embodiment indicated by reference numeral 11 in FIG. 1 or the standard pattern update promoting means 16 in the second embodiment indicated by reference numeral 16 in FIG. And a standard pattern registration switch 10
Of the writing patterns inputted by the writer through the input means 1 when the writer is turned off.
In the process of saving the writing pattern received by the data buffer 2 in the data buffer 2, the writer is urged to update the standard pattern from the saved number.

[0078] FIG. 14 is a normal operation in the author identification apparatus shown in FIG. 13, that is, a flowchart showing a flow of operation when the standard pattern registration switch 10 is OFF. In the figure, the part enclosed by the dotted line with reference numeral 17 corresponds to the operation of the standard pattern update promoting means 17. Note that steps ST360 to ST3
80, steps ST400 to ST410, step ST
20 to ST60, and step ST420, FIG.
The same processing as those steps in the first embodiment shown in FIG. Further, the operation at the time of the initial registration of the standard pattern in the writer identifying apparatus shown in FIG. 13, that is, the operation when the standard pattern registration switch 10 is ON, is the same as that of the first embodiment shown in FIG. Since they are the same, the description thereof will be omitted.

[0079] Next, the operation will be described. In the operations of steps ST360 to ST370 in FIG. 14, when the writing pattern input from input means 1 is accepted (step ST380), the accepted writing pattern is stored in data buffer 2 (step ST510). The standard pattern update promoting means 17 checks the number of writing patterns stored in the data buffer 2 and determines whether or not the number has reached a predetermined number N1 (step ST520).

[0080] Results of determination by the step ST 520, the writing pattern stored in the data buffer 2 is N1
If it is confirmed that the individual pieces exist, step ST400-
According to the processing of ST410, the screen shown in FIG. 8 is displayed on the display means 8 and the writer is asked whether or not to update the standard pattern registered in the writer identification dictionary 6. Only when the writer desires to update the writer identification dictionary 6, steps ST20 to ST60 and step ST4.
The operation of 20 is executed to update the standard pattern.
On the other hand, when the condition of ST520 is denied, the processing is ended as it is.

[0081] As described above, according to the third implementation, the standard pattern updating promoting means 17 first operates when the author identification means 7 the writing pattern input by the input means 1 was received, also it has been accepted Only the writing patterns are stored in the data buffer 2, and when the number of writing patterns received in the past stored in the data buffer 2 is collected in the number necessary for creating the standard pattern, the writing patterns are registered in the writer identification dictionary 6. Since the update of the standard pattern is urged, the confidentiality of the standard pattern of the writer himself / herself registered in the writer identification dictionary 6 is improved, the high security is maintained, and the most recently received writing is accepted. Since only the pattern is reflected in the standard pattern, the accuracy of the standard pattern of the writer registered in the writer identification dictionary 6 can be kept high. It has an effect .

In the third embodiment, the standard pattern registered in the writer identification dictionary 6 is updated by additionally registering the standard pattern. However, the same as in the first embodiment. As in the case of the second embodiment, the existing standard pattern may be deleted and rewritten with a newly created standard pattern.

[0083] Embodiment 4. 15 is a block diagram showing the configuration of the writer identification apparatus according to Embodiment 4 of the present invention, in which 1 to 11 in FIG.
Are the same as those in the first embodiment indicated by the same reference numerals. In addition, 400 surrounded by the one-dot chain line
Is a writing pattern verification means for determining whether or not to give permission to create a standard pattern to a writing pattern that the writer desires to register in the writer identification dictionary 6, and has a configuration as described below. It is a thing.

[0084] That is, in the writing pattern test means 400, 18 is recognized character recognition means what kind of character handwriting pattern writer is input from the input unit 1, the character recognition means 19 Reference numeral 18 is a character recognition dictionary in which characteristics of various characters necessary for recognizing the writing pattern into a certain character are described. Reference numeral 20 is a judging means for judging whether or not to give permission to create a standard pattern in order to newly register the standard pattern in the writer identification dictionary 6 based on the recognition result of the character recognition means 18.

[0085] FIG. 16 is the author identification apparatus shown in FIG. 15 is a flowchart showing a flow of an operation when registering a new standard pattern to the author identification dictionary 6.
In steps ST10 to ST80 in the figure, the same processes as those of the steps in the first embodiment shown in FIG. 2 are performed. Further, the operation at the normal time, that is, the operation when the standard pattern registration switch 10 is turned off is the same as that of the first embodiment shown in FIG. 7, and therefore the description thereof is omitted. It should be noted that the portion surrounded by the dotted line indicated by reference numeral 20 in FIG. 16 is the determination means 20 in FIG.
Corresponds to the processing of.

[0086] Next, the operation will be described. First, FIG.
After the standard pattern registration switch 10 is turned on in step ST10 in step ST10, in step ST530, the writing pattern of the character to be registered by the writer is entered. Thereafter, the writing pattern entered in step ST530 is input to the character recognition means 18 and character recognition is executed (step ST540). To execute this character recognition, for example, DP matching is performed between the writing pattern of characters input by the writer by the input means 1 and the standard pattern of each character registered in the character recognition dictionary 19 and the characters are input. Character recognition dictionary for writing patterns of characters 1
The dissimilarity Dc of each character registered in No. 9 with respect to the standard pattern is calculated. Then, among the differences Dc calculated for the standard patterns of each character registered in the character recognition dictionary 19, the character having the smallest value is selected, and this is used as the recognition result to end the processing. It is feasible.

[0087] It is to be noted that the construction of this character recognition dictionary 19,
It can be constructed by having multiple writers write the character pattern of some predetermined characters and registering the average of the sample point coordinates from the character pattern as a standard pattern. Is.

[0088] Then the recognition result of the character recognition means 18 is input to the determining means 20, then the resultant recognition result of the predetermined threshold dissimilarity Dc character TH_DIS
_C is compared (step ST550). If the character difference Dc of the recognition result exceeds the threshold value TH_DIS_C, the determination means 20 permits the creation of the standard pattern and continues the processing from step ST20. At this time, when there is no candidate for the character of the recognition result, that is, when it is determined to be rejected, the dissimilarity Dc indicates the maximum possible value. The processing after step ST20 is the same as that shown in FIG.
Since the operation is the same as that in the case of 1, the description thereof will be omitted. On the other hand, if the character difference Dc obtained from the recognition result is less than or equal to a predetermined threshold value TH_DIS_C, it is determined that the character to be registered does not have individuality, and the process proceeds to step ST560 to try to register. After outputting a message to the display means 8 indicating that the writing pattern of the existing characters does not have individuality, the standard pattern registration switch 10 is turned off in step ST80, and the process is terminated.

[0089] As described above, according to the fourth implementation, the handwritten pattern writer is about to be registered in the author identification dictionary 6, thereby executing the recognized beforehand by the character recognition unit 18, the recognition result character when dissimilarity is large only, namely the judgment unit 20 only when it is determined that the deviation from the standard pattern in the character recognition dictionary 19 large, since so as to allow the creation of a standard pattern, the author identification dictionary Only unique standard patterns will be registered in 6, and the accuracy of the original standard patterns must be maintained.
Can be, the effect there Ru to improve the security of the standard pattern of the writing in person.

[0090] Embodiment 5. 17 is a block diagram showing the configuration of the writer identification apparatus according to Embodiment 5 of the present invention, in which 1 to 11 in FIG.
Are the same as those in the first embodiment indicated by the same reference numerals. Also, 500 surrounded by the one-dot chain line
Is a writing pattern test means equivalent to the writing pattern test means in the fourth embodiment shown with reference numeral 400 in FIG. 15, and has a configuration as described below.

[0091] That is, in the writing pattern test means 500, 21 are registered frequency information table storing frequency information of the character registered in the author identification dictionary 6, 2
Reference numeral 2 indicates a frequency of use in the writer identification dictionary 6 for a desired character from the registration frequency information table 21, and when a new standard pattern is registered in the writer identification dictionary 6, the newly registered standard pattern is displayed. It is a registration frequency information searching / recording means for recording the usage frequency information of the characters constituting the above in the registration frequency information table 21. Reference numeral 23 is a determination means for determining whether or not to permit the standard pattern creation based on the frequency of use of the desired character.

[0092] In addition, the registration frequency information table 21,
For example, it is configured as shown in FIG. In the figure, taking "kawa" in the registered character column as an example, the frequency of use of the character "kawa" in the author identification dictionary 6 is a.
This means that there are a maximum of a standard pattern that uses the character "kawa" in the number of times.

[0093] FIG. 18 is the author identification apparatus shown in FIG. 17 is a flowchart showing a flow of an operation when registering a new standard pattern to the author identification dictionary 6.
In steps ST10 to ST80 in the figure, the same processes as those of the steps in the first embodiment shown in FIG. 2 are performed. Further, the operation at the normal time, that is, the operation when the standard pattern registration switch 10 is turned off is the same as that of the first embodiment shown in FIG. 7, and therefore the description thereof is omitted. Note that the portion surrounded by the dotted line indicated by the reference numeral 23 in FIG. 18 is the determination means 23 in FIG.
Corresponds to the processing of.

[0094] Next, the operation will be described. First, FIG.
After turning on the standard pattern registration switch 10 in step ST10, the writer corresponds to the writing pattern to be registered in the writer identification dictionary 6 by the input means 1 from the software keyboard displayed on the display-integrated tablet, for example. Input characters (step ST57)
0). After that, the input character is input to the registration frequency information search / recording means 22, and the registration frequency information table 21 is used to search the use frequency information of the input character in the writer identification dictionary 6 (step ST580). ). Next, the usage frequency for the desired character retrieved in step STS580 is input to the determination means 23 and compared with a predetermined threshold value X (step ST590).
As a result, if the frequency of use is X times or less, the determination means 23 permits the creation of the standard pattern, and the step ST
The processing of 20 to ST70 is continued. The processes of steps ST20 to ST70 are the same as those in the case of FIG.

[0095] Then, a new registration of the standard pattern to the author identification dictionary 6 in step ST70 is, do actually performed was to confirm that (step ST610). When the standard pattern is newly registered, the usage frequency information is recorded in the registration frequency information table 21 for the character registered this time by the registration frequency information search / recording means 22 (step ST620). Then, in step ST80, the standard pattern registration switch 10 is turned off, and the process ends. In step ST610, the author identification dictionary 6
If it is not confirmed that the standard pattern is newly registered, the process directly proceeds to step ST80, the standard pattern registration switch 10 is turned off, and the process is terminated.

[0096] On the other hand, in step ST590, if the use frequency was greater than threshold X times predetermined, the character to be registered it is judged that there is no security process proceeds to step ST600. In step ST600, a message indicating that the character to be registered has no security is output to the display means 8, and then in step ST80, the standard pattern registration switch 10 is turned off, and the process ends.

[0097] As described above, according to the fifth implementation, a character corresponding to a handwritten pattern writer is about to be registered in the author identification dictionary 6, the character in advance by registration frequency information retrieval / recording unit 22 Check the usage frequency information of
Since the determination means 20 permits the creation of the standard pattern only for the characters that are rarely used in the writer identification dictionary 6, the writer identification dictionary 6 uses the standard pattern based on the characters that are rarely used. Only the registered information will be registered, which has the effect of improving the security of the standard pattern of the writer himself.

In the fifth embodiment, the character code is input by the software keyboard displayed on the display-integrated tablet.
The writing pattern of the character to be registered is input to the character recognition means 18 and the character recognition dictionary 1 described in the fourth embodiment.
Alternatively, the characters may be recognized by using 9, and the characters may be coded and input.

In addition, in the fifth embodiment, it is determined whether or not to permit the creation of the standard pattern by using the registration frequency information table 21 that stores the registration frequency information of the characters in the writer identification dictionary 6. , The statistical information of characters used for first and last name, etc. may be used as the determination material.

[0100]

As described above, according to the present invention, the input writing pattern is compared with the standard pattern registered in the writer identification dictionary to determine the authenticity of the writing pattern, and the writing pattern is determined. When it is determined that the writer is himself, the step of creating a standard pattern based on the information related to the writing pattern and the created standard pattern are compared with the contents of the writer identification dictionary, and and a step of performing determining the validity of the update of the author identification dictionary, the updating of the author identification dictionary by the standard pattern when it is determined to be valid, the step of creating a standard pattern, the author identification dictionary To new
Because comprising the step of determining whether or not to grant the creation permission standard patterns for the writing pattern registration in the author identification dictionary is desired from the inputted handwritten pattern when registering a standard pattern, writing Standard pattern of the person
Security was maintained and there was a request from the scribe
It will be carried out only when
I don't need it. In addition, it is possible to create highly accurate standard patterns.
The effect is that the accuracy of the original standard pattern can be maintained.

According to the present invention, the writer identification dictionary for registering the standard pattern of handwriting, the input means for allowing the writer to input the writing pattern, the writing pattern of the writer and the standard pattern registered in the writer identification dictionary are provided. A writer identifying means for comparing the writing pattern to determine the authenticity, and a standard pattern creating means for creating a standard pattern from the writing pattern,
The writer identification dictionary is provided with a standard pattern registration / update determination means for registering or updating the standard pattern created by the standard pattern creating means based on the collation result of the standard pattern created by the standard pattern creating means and the contents of the writer identification dictionary.
From the writing patterns input by the writer by the writer when a new standard pattern is newly registered to, whether the writing writer wants to create a standard pattern for the writing pattern that the writer desires to register in the writer identification dictionary. Since a writing pattern verification means for judging whether or not it has been established, the standard pattern of the writer himself
The turn security is maintained, and the demand from the scribe
It will be executed only when there is a wasteful process
It doesn't have to be done. In addition, highly accurate standard pattern creation
Adult can, the effect of I identification device is obtained which can maintain the accuracy of the original standard pattern.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a writer identification device according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing an operation at the time of initial registration of a standard pattern according to the first embodiment.

FIG. 3 is a flowchart showing the operation of the pattern selection means in the first embodiment.

FIG. 4 is a flowchart showing details of the operation of step ST160 in FIG.

FIG. 5 is a flowchart showing details of the operation of step ST190 in FIG.

FIG. 6 is a flowchart showing an operation of standard pattern registration / update determination means in the first embodiment.

FIG. 7 is a flowchart showing a normal operation in the first embodiment.

8 is an explanatory diagram showing a display example of a message displayed on the display means in step ST400 in FIG. 7. FIG.

FIG. 9 is a flowchart showing the operation of the writer identification means in the first embodiment.

FIG. 10 is an explanatory diagram showing the contents of a writer identification dictionary according to the first embodiment.

FIG. 11 is a block diagram showing a configuration of a writer identification device according to a second embodiment of the present invention.

FIG. 12 is a flowchart showing a normal operation according to the second embodiment.

FIG. 13 is a block diagram showing a configuration of a writer identification device according to a third embodiment of the present invention.

FIG. 14 is a flowchart showing a normal operation according to the third embodiment.

FIG. 15 is a block diagram showing a configuration of a writer identification device according to a fourth embodiment of the present invention.

FIG. 16 is a flowchart showing an operation at the time of initial registration of a standard pattern in the fourth embodiment.

FIG. 17 is a block diagram showing a configuration of a writer identification device according to a fifth embodiment of the present invention.

FIG. 18 is a flowchart showing an operation at the time of initial registration of a standard pattern according to the fifth embodiment.

FIG. 19 is an explanatory diagram showing the contents of a registration frequency information table 21 according to the fifth embodiment.

FIG. 20 is a flowchart showing an operation of initial registration of a standard pattern in conventional writer identification.

FIG. 21 is a flowchart showing an operation of updating a standard pattern in conventional writer identification.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 input means, 2 data buffer, 3 pattern selection means, 4 standard pattern creation means, 5 standard pattern registration / update determination means, 6 writer identification dictionary, 7 writer identification means, 11, 16, 17 standard pattern update promotion means, 1
8 character recognition means, 19 character recognition dictionary, 20, 23
Judgment means, 21 Registration frequency information table, 22 Registration frequency information search / recording means, 400, 500 Writing pattern verification means.

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Yuichi Okano 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Mitsubishi Electric Corporation (72) Inventor Isamu Ogawa 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Mitsubishi Electric (56) References JP 62-177680 (JP, A) JP 2-89195 (JP, A) JP 7-57093 (JP, A) JP 4-320583 (JP, A) A) JP-A-7-141506 (JP, A) JP-A-7-29001 (JP, A) JP-A-7-121663 (JP, A) JP-A-7-225818 (JP, A) JP-A-60 -72067 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T ⁷ /00-7/60 G06T 1/00

Claims (7)

(57) [Claims] 1. The authenticity of the writing pattern is judged by comparing the input writing pattern with a standard pattern registered in the writer identification dictionary, and the writing pattern is judged to belong to the writer himself. , A step of creating a standard pattern based on the information related to the writing pattern and the created standard pattern are collated with the contents of the writer identification dictionary to determine the validity of updating the writer identification dictionary by the standard pattern. And a step of updating the writer identification dictionary with the standard pattern when it is determined to be appropriate, in the step of creating the standard pattern, the writer identification
When registering a new standard pattern in the dictionary, a step of determining whether or not to grant the creation of the standard pattern to the writing pattern desired to be registered in the writer identification dictionary from among the input writing patterns A method for identifying an author, comprising: 2. A writer identification dictionary for registering a standard pattern of handwriting, an input means for allowing a writer to input a writing pattern, and a comparison of the writing pattern of the writer and a standard pattern registered in the writer identification dictionary. And a writer identification means for determining the authenticity of the writing pattern, a standard pattern creating means for creating a standard pattern from the writing pattern,
And a standard pattern registration / update determination means for registering or updating the standard pattern created by the standard pattern creation means based on the collation result of the standard pattern created by the standard pattern creation means and the content of the writer identification dictionary. In the writer identification device, when newly registering a standard pattern in the writer identification dictionary, the writer desires to register in the writer identification dictionary from among the writing patterns input by the writer by the input means. An writer identification device, characterized by comprising a writing pattern verification means for judging whether or not to give permission to create a standard pattern for the writing pattern. 3. The writing pattern verification means stores a character recognition dictionary in which a unique average pattern is registered for each character and an average pattern for each character registered in the character recognition dictionary. Character recognition means for recognizing what kind of character the writing pattern is, as compared with the writing pattern, and determination for giving permission to create a standard pattern for the writing pattern based on the recognition result of the character recognition means. The writer identification device according to claim 2, further comprising: 4. The writing pattern verification means registers registration of usage frequency information indicating what kind of character a standard pattern registered in the writer identification dictionary is and the usage frequency of the character in the writer identification dictionary. Frequency information table, searching the use frequency information of the character code corresponding to the writing pattern by referring to the registration frequency information table, when the standard pattern for the writing pattern is registered,
Based on the registration frequency information search / recording unit that records the use frequency information of the character code corresponding to the writing pattern in the registration frequency information table, and the use frequency information of the character code searched by the registration frequency information search / recording unit. 3. The writer identifying apparatus according to claim 2, further comprising: a determining unit that gives permission to create a standard pattern for the writing pattern. 5. A pattern selection means for selecting a writing pattern effective for creating a standard pattern from among the writing patterns input from the input means based on a stroke of the shape of the writing pattern. Claim 2
The described writer identification device. 6. The pattern selecting means determines the complexity of a shape based on a ratio of a stroke length forming a writing pattern and a linear distance between the start and end points of the stroke. Author identification device. 7. The writer identification device according to claim 5, wherein the pattern selection means determines the complexity of the shape based on the direction information between the sample points of the strokes forming the writing pattern. .