JPH01204196A - Method and device for recognizing continuous numeral and its device - Google Patents

Abstract

PURPOSE:To improve the recognizing processing speed of a continuous numeral by preparing a continuous numeral pattern having the assumed number of digits in the higher sequence of its generation probability and collating with an input pattern sequentially from the continuous numeral pattern with a high generation probability. CONSTITUTION:A digit number assuming means 12 assumes the number of the digits of an input numeral based on the characteristic of an input pattern, in a continuous numeral pattern preparing means 14, the figure continuous probability of the continuous numeral having the number of assumed digits instructed from the digit number assuming means 12 based on the numeral continuous probability of a figure continuous probability storing part 16 is obtained, the numeral pattern is removed from a numeral pattern dictionary part 13 and the continuous numeral pattern having the assumed number of the digits is prepared in the generation probability sequence. In a collating means 15, the prepared continuous numeral pattern is collated with the input pattern in the higher sequence of the generation probability, and the input pattern and the continuous numeral pattern with the highest approximation are outputted as the recognized result. Thus, the recognition processing speed of a continuous numeral can be improved.

Description

[Detailed Description of the Invention] [Summary] Concerning a continuous number recognition method and device that recognizes input continuous numbers by comparing an input continuous number pattern with a synthesized continuous number pattern, the recognition rate is reduced with a small number of consecutive number patterns. The purpose is to speed up the continuous number recognition process without causing the input pattern to be distorted.The system estimates the number of digits of the input number from the input pattern, generates a continuous number pattern with the estimated number of digits from the individual number patterns, and then combines the input pattern with the continuous number pattern. In a continuous number recognition method and device that recognizes input consecutive numbers by comparing them, continuous number patterns having an estimated number of digits are generated based on the probability of consecutive numbers in descending order of probability of occurrence, and continuous number patterns with the highest probability of occurrence are generated. Configure it to match input patterns in order. Further, the consecutive number probability is configured to be changeable in accordance with the writing order of the numbers.

[Industrial application field]

The present invention relates to a continuous number recognition method and apparatus for recognizing consecutive numbers written in Zokuha characters, and in particular, to recognize input consecutive numbers by comparing a continuous number pattern synthesized from an input consecutive number pattern and an individual number pattern. The present invention relates to a continuous number recognition method and device.

[Conventional technology]

In the field of online character recognition, characters written individually within a certain frame have traditionally been recognized, but in recent years, in order to realize a freer handwriting environment, the recognition of characters without a frame has been performed. It has become common practice to recognize characters written in the original language.

When there are no restrictions on the entry frame and it becomes possible to freely input handwritten characters, input characters may become continuous characters. In particular, for numbers, the way each character is written is simple, so if there is no frame, for example, if you write numbers consecutively, such as when writing numbers on a slip, the continuation will be the consecutive numbers of the characters. There is a strong tendency to

FIG. 3 is a block diagram showing the configuration of a continuous number recognition device that recognizes such consecutive numbers written in Japanese characters.

In FIG. 3, reference numeral 31 denotes an input unit, which is composed of a tablet or a digitizer, and when numbers (including consecutive numbers) are input by hand, each coordinate data of the input numbers is output.

32 is an input pattern generation section that generates an input pattern based on input numerical coordinate data from the input section 31;

A digit number estimation unit 33 estimates the number of digits of an input number from characteristics such as the horizontal length of the input pattern.

34 is a number pattern dictionary section in which each number pattern is stored for each individual number.

Reference numeral 35 denotes a continuous number synthesis section which extracts individual number patterns for the estimated number of digits specified by the number of digits estimation section 33 from the number pattern dictionary section 34, and synthesizes all consecutive numbers having this specified number of digits.

36 is a continuous number pattern generation unit, and a continuous number synthesis unit 35
Generates a continuous digit pattern for each consecutive digit combined with .

37 is a matching unit that matches the input pattern generated by the input pattern generation unit 32 and the continuous number pattern generated by the continuous number pattern generation unit 36, and generates a continuous number pattern having a degree of approximation equal to or higher than a predetermined degree of approximation value. Select and output candidate consecutive numbers.

Reference numeral 38 denotes a recognition result output unit, which outputs the matching result inputted from the matching unit 37 with the highest degree of approximation among candidate consecutive digits as a recognition result.

With this configuration, when consecutive numbers are input by hand to the input unit 31, each coordinate data of the consecutive numbers is sent to the input pattern generation unit 32, and an input pattern of consecutive numbers is generated.

The number of digits estimating unit 33 estimates the number of digits of the input consecutive numbers based on the characteristics of the generated input pattern, and instructs the consecutive number synthesizing unit 35 to estimate the number of digits of the input consecutive numbers.

The continuous number synthesis unit 35 extracts the number pattern for the estimated number of digits instructed by the number of digits estimation unit 33 from the dictionary unit 34,
Combine all consecutive numbers with this estimated number of digits.

The continuous pattern generation unit 36 generates a continuous Via 1-character pattern based on the data of each continuous number synthesized by the continuous number synthesis unit 35. Table 1 shows the number of consecutive number digits and the number of consecutive number patterns generated. In the case of a 9-digit number, 10'' consecutive number patterns are generated.

Table 1 The matching unit 37 matches the continuous number input pattern generated by the input pattern generation unit 32 with the continuous number pattern generated by the continuous pattern generation unit 36, and determines the degree of approximation equal to or higher than a predetermined degree of approximation value. Extract a continuous number pattern having , and output candidate consecutive numbers.

The recognition result output unit 38 outputs the one with the highest degree of approximation (recognition rate) from among the candidate consecutive digits inputted by the matching unit 37 as a recognition result.

The matching unit 37 restarts the digit number estimating unit 33 when the degree of approximation of the candidate consecutive digits with the highest matching degree is less than a predetermined darkness value. The number of digits estimating unit 33 re-estimates the number of digits of the input number string and instructs the continuous number synthesizing unit 35 to estimate the number of digits of the input number string. Thereafter, the aforementioned processes are repeated for this newly re-estimated number of digits, and the continuous number pattern with the highest degree of approximation (recognition rate) is output as a recognition result.

Note that if the input number is one digit, the processing by the continuous number synthesis section 35 and the continuous number pattern generation section 36 is not necessary.
The numeric pattern taken out from the numeric pattern dictionary section 34 passes through both sections 35 and 36 as is, and is sent to the matching section 37, where it is matched against the input pattern.

[Problem to be solved by the invention]

In the conventional continuous number recognition method, as described above, consecutive numbers are synthesized from all number strings with an estimated number of digits of input consecutive numbers, and the consecutive number patterns are compared with the input pattern.

The number of consecutive digits consisting of any sequence of numbers in the estimated arithmetic number is
It is proportional to the square of the estimated number of digits n. For this reason, when the number of lit digits reaches a fixed number of digits, the number of consecutive number patterns generated becomes extremely large, as shown in Table 1, and there is a problem in that the processing speed of consecutive numbers decreases significantly.

Even if the estimated number of consecutive digits in the input sequence increases, the present invention effectively suppresses the increase in the number of generated consecutive number patterns that are matched with the input pattern, and performs consecutive number recognition processing without reducing the recognition rate. It is an object of the present invention to provide a method and apparatus for recognizing consecutive numbers that are improved so as to be faster.

[Means to solve the problem]

When writing a string of numbers, depending on the number, the continuation tends to be a letter (for example, numbers 2, 5, 8, etc.).
0 etc. may continue from the back as well as from the front. On the other hand, the numbers 1, 3, 7.9
etc. may continue from the previous character, but it is unlikely to continue from the next character. In other words, the probability of consecutive numbers differs depending on each number.

From this, even if the number of digits is the same, the probability of consecutive digits differs depending on the type of digit, so the probability of occurrence will be different. Therefore, if consecutive number patterns with a high probability of occurrence are generated preferentially and matched with the input pattern, the recognition process can be efficiently performed with a good recognition rate and with a small number of consecutive number pattern generation. It can be carried out.

Each invention of the present application was made based on such an idea, and the structure of each invention will be explained separately into the structure of the method invention and the structure of the device invention used to implement the method invention.

(1) Structure of the method invention The method invention estimates the number of digits of an input number from an input pattern, generates a continuous number pattern having the estimated number of digits from the individual number patterns, and compares the input pattern with the continuous number pattern. In a continuous digit recognition method that recognizes input consecutive digits, a continuous digit pattern with an estimated number of digits is generated with a high probability of occurrence based on the probability of consecutive digits, and the consecutive digit patterns are input in order from the highest probability of occurrence. Configured to match a pattern.

The probability of consecutive numbers of any number of digits can be determined from the probability that any two numbers (including the case where they overlap) are consecutive. For this reason, another method invention is configured to obtain the probability of consecutive numbers having an estimated number of digits from the probability of consecutive numbers of two numbers.

Even if the probability of successive numbers after one number and the probability of successive numbers following the previous number are determined using high/low probability information indicating the two values of high and low, the recognition rate decreases. Recognition processing can be performed efficiently without any problems.

For this reason, still another method invention uses high and low probability information that indicates the probability that a number will be consecutive after one number and the probability that that number will be consecutive to the previous number as binary values of high and low. configured to set.

These high/low probability information are determined, for example, by the following procedure.

■ If the end point of the last stroke of a number is on the rightmost side of the number, and the direction of the last part of the number is from the top to the right, increase the probability that it will continue with the next number. In other cases, the probability of successive numbers with the next number is lowered.
・・・・・・・・・Procedure ■■ When the starting point of the first stroke of a number is not located at the leftmost position of the whole number, the probability that it is continuous with the previous number is lowered. In other cases, the probability of consecutive numbers with the previous number is increased. ...Procedure (2) Therefore, still another method invention is configured to determine each of the high and low probability information according to the above-mentioned procedures (2) and (2).

The probability of consecutive numbers may change depending on how they are written. For this reason, yet another method invention is configured such that the number consecutive probability of consecutive numbers can be changed in response to different stroke orders.

(2) Structure of device invention The configuration of the device invention will be explained with reference to FIG.

FIG. 1 is a block diagram showing the basic configuration of one consecutive digit recognition device used to implement the method invention according to the present invention.

In FIG. 1, reference numeral 11 denotes an input pattern generation means that generates an input pattern of input numbers.

Reference numeral 12 denotes a digit number estimating means, which performs a process of estimating the number of digits of an input number based on the characteristics of the input pattern.

13 is a number pattern dictionary section in which patterns of individual numbers are stored.

14 is a continuous number pattern generating means, which calculates the probability of occurrence of consecutive numbers having the estimated number of digits instructed by the number of digits estimating means 12 based on the number consecutive probability stored in the number consecutive probability storage section 16, and calculates the probability of occurrence of consecutive numbers having the estimated number of digits instructed by the number of digits estimating means 12, A number pattern is extracted and consecutive number patterns having an estimated number of digits are generated in order of their probability of occurrence.

Reference numeral 15 denotes a collation means, which collates the continuous number patterns generated by the continuous number pattern generation means 14 with input patterns in descending order of probability of occurrence, and outputs the continuous number pattern with the highest degree of approximation to the input pattern as a recognition result.

The number consecutive probability storage unit 16 stores the number consecutive probability of consecutive numbers. In this case, given the probability of two consecutive numbers among the ten numbers 0 to 9, including duplicate cases, the probability of consecutive numbers of any number of digits can be found, so By storing only two-number consecutive probabilities for ten numbers, the capacity and configuration of the number consecutive probability storage section 16 can be simplified (details will be explained in the section of the embodiment).

17 is a number consecutive probability changing means, which is a number consecutive probability storage unit 1;
Processing is performed to change the number consecutive probability stored in 6. In the case of handwritten numbers, the same number may be written differently depending on the person (for example, the numbers "4" and "5"). If a number is written in different ways, the probability of occurrence of consecutive numbers including that number will differ depending on how it is written. If the probability of occurrence of consecutive numbers differs depending on the writing order as described above, the number consecutive probability changing means 17 changes the number consecutive probability stored in the number consecutive probability storage unit 16 in accordance with the writing order of the input digits ( Details are explained in the Examples section).

[For production]

(1) Effect of the method invention Each method invention generates consecutive number patterns having an estimated number of digits in order of their probability of occurrence, and compares them with the input pattern in order of successive number patterns having the highest probability of occurrence.
The continuous number pattern with the highest degree of approximation is output as the recognition result.

As a result, it is possible to recognize input consecutive digits with fewer consecutive digit patterns generated than in the conventional method, and therefore it is possible to improve the recognition processing speed of consecutive digits with a good recognition rate.

Further, by determining the probability of consecutive numbers for consecutive numbers having an estimated number of digits from the probability of consecutive numbers for two numbers, the configuration of the consecutive numbers probability storage unit 16 can be simplified.

Furthermore, by determining the successive probability of two numbers using high-low probability information that indicates the probability that a number will be consecutive after one number and the probability that that number will be consecutive to the previous number as binary values of high and low, The configuration of the number consecutive probability storage section 16 can be further simplified, and the continuous number pattern generation process can be sped up.

By making it possible to change the probability of occurrence of consecutive digits in response to different stroke orders, it is possible to improve the recognition rate of input consecutive digits and speed up the recognition process of consecutive digits.

(2) The action input pattern generating means 11 of the device invention generates an input pattern of input numbers.

The number of digits estimating means 12 estimates the number of digits of the input number based on the characteristics of the input pattern, and instructs the continuous number pattern generating means.

The consecutive number pattern generation means 14 calculates the number consecutive probability of consecutive numbers having the estimated number of digits instructed by the number of digits estimating means 12 based on the number consecutive probability stored in the number consecutive probability storage section 16, and calculates the number consecutive probability from the number pattern dictionary section 13. A number pattern is extracted and consecutive number patterns having an estimated number of digits are generated in order of their probability of occurrence.

The collation means 15 collates the continuous number patterns generated by the continuous number pattern generation means 14 with the input patterns in descending order of probability of occurrence, and outputs the continuous number pattern with the highest degree of approximation to the input pattern as a recognition result.

As a result, input consecutive digits can be recognized with fewer consecutive digit patterns generated than in the conventional method, and the recognition processing speed of consecutive digits can be improved with a good recognition rate.

In addition, for consecutive numbers, the successive probability of two numbers is stored in the growth rate storage unit 16, and the consecutive number pattern generating means 14 calculates the probability of consecutive numbers of consecutive numbers having an estimated number of digits based on the consecutive probability of two numbers. By doing so, the configuration of the number consecutive probability storage section 16 can be simplified.

Furthermore, as consecutive probability information for two numbers stored in the number consecutive probability storage unit 16, the 'α rate of successive numbers after one number and the probability that that number follows the previous number are expressed as high or low, respectively. By storing high/low probability information expressed in binary values, the configuration of the consecutive number probability storage section 16 can be further simplified, and the continuous number pattern generation process can be sped up.

By providing a digit sequence probability changing means 17 that changes the digit sequence probability stored in the digit sequence probability storage unit 16 in accordance with the writing order of input digits, the recognition rate of input sequence digits is improved, and continuous digit recognition is improved. Processing can be sped up.

〔Example〕

Examples of each invention of the present application will be described with reference to FIG. 2. FIG. 2 is a block diagram showing the configuration of a continuous digit recognition device used to implement the method invention of the present application.

(A) Structure of continuous digit recognition device In FIG. Number pattern dictionary section 13. Continuous VC number pattern generation means 14. Number consecutive probability storage unit 16. The number consecutive probability changing means 17 is as described in FIG.

In the input pattern generation means 11, reference numeral 111 denotes an input unit, which is composed of a tablet or a digitizer, and when numbers (including consecutive numbers) are input by hand, each coordinate data of the input numbers is output. Reference numeral 112 denotes an input pattern generation section that generates an input pattern based on the input numerical coordinate data from the input section 111.

In the continuous number pattern generating means 14, 141 is a continuous number synthesizing section, which refers to the number consecutive probability storage section 16 to calculate the number consecutive probability of consecutive numbers having the estimated number of digits instructed by the number of digits estimating means 12. A number pattern is taken out from the pattern dictionary section 13, and consecutive numbers having an estimated number of digits are synthesized in order of their probability of occurrence. 142 is a continuous number pattern generation unit that generates continuous number patterns having an estimated number of digits in the order of their probability of occurrence based on the data of each continuous number synthesized by the continuous number synthesis unit 141.

In the matching means 15, 151 is a matching unit that matches the input pattern generated by the input pattern generation unit 112 and the continuous number pattern generated by the continuous number pattern generation unit 142, and obtains a degree of approximation equal to or higher than a predetermined degree of approximation value. Select a consecutive number pattern having the same number and output candidate consecutive numbers.

Reference numeral 152 denotes a recognition result output unit, which outputs the one with the highest degree of approximation from among the candidate consecutive numbers inputted from the matching unit 151 as a recognition result.

In the number consecutive probability changing means 17, 171 is a test character input section, which is composed of a tablet, a digitizer, etc.
When a test character is input to determine the order in which a number can be written in multiple ways, the test character coordinate data is output. 172 is a stroke order detection unit that identifies the stroke order of each number, which can be written in a plurality of ways, based on the test sentence coordinate data, and outputs the identification result. Numeral 173 is a number consecutive probability changing unit which changes the number consecutive probability of consecutive numbers stored in the number consecutive probability storage unit 16 based on the identification result.

(B) Operation of the continuous digit recognition device Operation of the continuous digit recognition device, ie, a method Embodiments of the invention include:
The number consecutive probability stored in the number consecutive probability storage unit 16 is 1.
An example will be explained in which the probability of successive numbers before and after one number is the successive probability of two numbers given by high/low probability information.

First, the two-number consecutive probability consisting of high-low probability information stored in the consecutive number probability storage unit 16 will be explained.

For each number from 0 to 9, whether the probability of successive numbers (including the same number) before and after each number is high or low is determined by the steps ① and ② explained earlier, according to many experiments. be done. In other words, step ■: If the end point of the last stroke of a number is on the rightmost side of the number, and the direction of the last part of the number is between the top and the right, increase the probability that it will continue with the next number. do. In other cases, the probability of successive numbers with the next number is lowered.

Procedure ■: If the starting point of the first stroke of a number is not located at the leftmost position of the entire number, lower the probability that it will be continuous with the previous number. In other cases, the probability of consecutive numbers with the previous number is set to high.

Table 2 shows the numbers O~ determined by these procedures ■ and ■.
This shows the probability of successive numbers before and after each number 9.

In Table 2, column [a] shows numbers O to 9. Column (b) shows the order in which each number is written. (Column C1 indicates the direction of the line of the last stroke of each number. (d) l[1j indicates the probability of each number being consecutive to the next number, and “○” indicates a high probability. "X" indicates a low probability. Column (e) indicates whether the starting point of the first stroke of each number is the leftmost position of the number, and "Y" indicates an affirmative. “N”
indicates negation. Column (f) shows the probability of successive numbers following the previous number; “○” indicates a high probability; “
×” indicates low.

As can be seen from Table 2, for numbers 4 and 5, 41) and 4(, ) and 5 +01 and 5(L, respectively) are 2
There is a stroke order of A, and depending on the stroke order, numbers 4 and 5 are written.
The probability of consecutive numbers before and after changes. The order in which the number 4 is written is determined by the following procedure (2), and the order in which the number 5 is written is determined by the following procedure (2).

Procedure ①: For the number 4, compare the X coordinates of the end points of the two strokes "mu" and "1", and determine that the one with the larger X coordinate is the stroke "mu".

Procedure ①: For the number 5, compare the X coordinates of the end points of the two strokes "5" and "-", and determine that the one with the larger X coordinate is the stroke "-".

To summarize Table 2, as shown in Table 3, the number 2.4+t
For +, 5 (U), 8 and 0, the probability or possibility of continuing with the following number is large, and the numbers 1, 3゜4 (U),
5n+, 6.7 and 9 have small probabilities or possibilities. On the other hand, the number m2.3.4yu.

5 (U, , 6, 7, 8.9 and O have a large probability or possibility of continuing to the previous number, and numbers 4 (L) and 5 (L
The probability, or possibility, of l is small.

Table 3 In the following explanation, the number consecutive probability storage unit 16 includes:
The two-number consecutive probabilities according to Table 2 (same as Table 3) are stored, but numbers 4 and 5 are determined by testing the actual writing order.

Next, the operation of the continuous number recognition device will be explained in the order of processing. First, as pre-processing, the two-number consecutive probability stored in the consecutive number probability storage unit 16 is changed.

■ Numbers that can be written in multiple ways (numbers 4 and 5 in the example)
), specify the numbers (4 and 5) whose stroke order needs to be determined to the scribe of the number to be recognized,
The test character input section 171 is caused to input handwriting.

(2) When the test characters are input by hand, the test character input section 171 sends digitized signals of the coordinate data of the input digits 4 and 5 to the stroke order detection section 172.

■ The stroke order detection unit 172 identifies the four-way order of the numbers 4 and 5 from the coordinate data of the input numbers 4 and 5 according to the above-mentioned steps ■ and ■, and applies the identification result to the number consecutive probability changing unit 173. send to

(2) Upon receiving this identification result, the number consecutive probability changing section 173 changes the probability values of numbers 4 and 5 among the two number consecutive probabilities stored in the number consecutive probability storage section 16 according to Table 2.

After the above preprocessing is completed, recognition processing is performed on the actual handwritten input digits.

(2) When numbers are input by hand to the input unit 111, the coordinate data is sent to the input pattern generation unit 112, and an input pattern of the input numbers is generated.

(2) The number of digits estimating means 12 estimates the number of digits of the input digits based on the characteristics of the input pattern, and instructs the continuous number synthesis section 141 to estimate the number of digits of the input digits.

■ The continuous number synthesis unit 141 is connected to the continuous number probability storage unit 16.
Find the probability of occurrence of consecutive numbers using the estimated number of digits specified. The one with the highest probability of two consecutive numbers (“○” in Table 2)
PL the value of the lower one (marked with “x” in Table 2)
Then, the probability of one consecutive digit having an estimated number of digits n is obtained as the product of these P and PL.

In the case of Table 2, there are multiple number strings with an estimated number of digits n that have the same number sequence probability. The consecutive number synthesis unit 141 first extracts each number pattern of each number string having the estimated number of digits n with the highest probability of consecutive numbers from the number pattern dictionary unit 13, synthesizes the consecutive numbers having the estimated number of digits n, and then It is sent to the number pattern generation section 142.

(2) The continuous number pattern generation unit 142 generates a continuous number pattern having an estimated number of digits n based on the input data of each continuous number.

■ The collation unit 151 collates each input consecutive number pattern with the input pattern generated by the input unit pattern generation unit 112, selects a consecutive number pattern having a degree of approximation equal to or higher than a predetermined darkness value, and determines the number of candidate consecutive numbers. as recognition result output unit 15
Send to 2.

If there is no continuous number pattern with a degree of approximation equal to or higher than the predetermined darkness value, the following process a or b is performed depending on the configuration of the device.
I do.

Processing ■ The collation unit 151 restarts the continuous number synthesis unit 141. Continuous number synthesis section 141 and continuous number pattern generation section 142
is stored in the number consecutive probability storage unit 1 in the same way as in the above ■~■
6 and the number pattern and the dictionary unit 13 to generate a continuous number pattern with the estimated number of digits n that has the next highest probability of occurrence, and the matching unit 15
Send to 1.

Processing ■ The matching unit 151 restarts the digit number estimating means 12. The number of digits estimating means 12 re-estimates the number of digits of the input pattern that has been input, and sends it to the continuous number synthesis section 151.

If the precision of the number of digits estimating means 12 is high, the process (2) is performed, and if the precision of the number of digits estimating means 12 is low, the process (2) is performed. In some cases, image processing is configured to be performed alternately.

As described above, the matching unit 151 selects the continuous number pattern with the estimated number of digits n, which has the highest probability of occurrence with a degree of approximation equal to or higher than the dark value, and is selected by the recognition result output unit 152 as a candidate continuous number.
sent to.

[Phase] The recognition result output unit 152 outputs the one with the highest degree of approximation from among the input candidate consecutive numbers as a recognition result.

Now, the writing order of numbers 4 and 5 in the numbers to be recognized is 4 (U
) and 5 +u+, the two-number consecutive probabilities in Table 3 can be summarized as shown in Table 4 below.

Table 4 Table 5 shows the ratio between the number of consecutive number patterns of each number of digits that are preferentially generated in this case and the number of consecutive patterns that are generated in the conventional device (see Table 1).

Table 5 From this, the continuous pattern that is preferentially generated for 0 digits is
Assuming that the probability of occurrence in an actual input pattern is P7, the ratio between the number of consecutive digits actually generated and the number in the case of the conventional device is expressed by the following equation.

(Number of consecutive numbers generated by the device of the present invention (expected value))/(
Number of consecutive numbers generated by conventional equipment) = (0, 4)”-
' xp, , + (1-P, 1) As described above, according to the present invention, it is possible to efficiently perform recognition processing with a good recognition rate while generating far fewer consecutive number patterns than conventional devices. can.

In addition, if the number of digits of the input number is 1, continuous number synthesis section 1
41 and continuous number pattern 142 are not required;
Each number pattern taken out from the number pattern dictionary section 13 is sent as is to the matching section 151.

Although the embodiments of the method and apparatus of the present invention have been described above, the embodiments of the method and apparatus of the present invention are not limited to these embodiments.

The test character input section 171 and the input section 111 can be configured as a common input section. Furthermore, instead of detecting the stroke order, it is possible to show the scribe the stroke order of numbers in a plurality of stroke orders and allow the scribe to select one.

〔Effect of the invention〕

As explained above, according to each invention of the present application, the following effects can be obtained.

The number of consecutive inputs can be recognized with fewer consecutive number patterns generated than in the conventional method, and the recognition processing speed of consecutive numbers can be improved with a good recognition rate.

By determining consecutive numbers having an estimated number of digits from the consecutive probability of two numbers, the configuration of the number consecutive probability storage unit can be simplified.

2 using the information on the probability that a number will be consecutive after one number and the information on the probability that the number will be consecutive to the previous number.
By determining the probability of successive numbers each time, the configuration of the successive number probability storage section can be simplified, and the continuous number pattern generation process can be sped up.

By making it possible to change the probability of occurrence of consecutive digits in accordance with the writing order of input digits, it is possible to improve the recognition rate of input consecutive digits and speed up the recognition process of consecutive digits.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram of the basic configuration of a continuous digit recognition device used to carry out the method invention of the present application. FIG. 2 is an explanatory diagram of the configuration of an embodiment of the same continuous digit recognition device. , is an explanatory diagram of the configuration of a conventional continuous number recognition device. 1 and 2, 11...input means, 111...input section, 112...
- Input pattern generation unit, 12... digit number estimation means, 13
...Number pattern dictionary section, 14...Continuous number pattern generation means, 141...Continuous number synthesis section, 142...
・Continuous number pattern generation unit, 15... Collation means, 15
1... Verification unit, 152... Recognition result output unit, 16.
. . . Number continuity probability storage unit, 17 . . . Number continuity probability changing unit, 171 . . . Test character input unit, 172 . Fig. 1 Basic configuration of continuous digit recognition device used to implement the method invention

Claims (1)

[Claims] 1. Estimate the number of digits of input numbers from the input pattern, generate a continuous number pattern having the estimated number of digits from the individual number patterns, and compare the input pattern with the continuous number pattern to determine the number of input numbers. In a continuous number recognition method that recognizes consecutive numbers, (A) consecutive number patterns having an estimated number of digits are generated based on the probability of consecutive numbers in descending order of probability of occurrence, and (B) consecutive number patterns with a high probability of occurrence are generated. A continuous number recognition method characterized by sequentially matching input patterns. 2. The method according to claim 1, characterized in that the successive probability of consecutive numbers having an estimated number of digits is determined from the consecutive probability of two numbers, and consecutive number patterns having an estimated number of digits are generated in descending order of probability of occurrence. Continuous number recognition method. 3. The consecutive number recognition method according to claim 2, wherein the consecutive number probability is comprised of a consecutive probability for a previous number and a consecutive probability for a subsequent number. 4. A feature is that the probability of two consecutive numbers is set using high/low probability information that indicates the probability that a number will be consecutive after one number and the probability that that number will be consecutive to the previous number, respectively, using binary values of high and low. The continuous number recognition method according to claim 2. 5. The continuous number recognition method according to claim 4, wherein the high/low probability information is determined by the following procedure. [1] If the end point of the last stroke of a number is on the rightmost side of the number, and the direction of the last part of the number is from the top to the right, the probability of successive strokes with the next number is increased. In other cases, the probability of successive numbers with the next number is lowered. [2] When the starting point of the first stroke of a number is not located at the leftmost position of the entire number, the probability that it continues with the previous number is lowered. In other cases, the probability of consecutive numbers with the previous number is increased. 6. Claim 2, 4 or 5, characterized in that the probability of occurrence of consecutive digits can be changed in accordance with different writing orders of the digits.
Continuous number recognition method described in section. 7. Input pattern generation means for inputting consecutive numbers (11
), a digit number estimation means (12) for estimating the number of digits of an input number based on the characteristics of the input pattern, a number pattern dictionary section (13) in which individual number patterns are stored, and a number pattern dictionary section (13). ), the continuous number pattern generating means (14) extracts a number pattern from the input pattern and generates a continuous number pattern having an estimated number of digits, and the matching means (15) matches the continuous number pattern with the input pattern and outputs a recognition result. In the consecutive number recognition device, (A) a number consecutive probability storage section (16) is provided in which the number consecutive probability of consecutive numbers is stored, and (B) the consecutive number pattern generation means (14) is provided with a number consecutive number probability storage section ( 16) Find the probability of occurrence of consecutive numbers having an estimated number of digits based on the number of consecutive numbers, extract the number patterns from the number pattern dictionary section (13), and generate consecutive number patterns having the estimated number of digits in the order of their probability of occurrence. , (C
) A continuous number recognition device characterized in that the matching means (15) matches consecutive number patterns with input patterns in descending order of probability of occurrence. 8. The continuous number recognition device according to claim 7, wherein the number consecutive probability stored in the number consecutive probability storage section (16) is a number consecutive probability for two numbers. 9. The consecutive digit recognition device according to claim 8, wherein the two-digit consecutive probability consists of a consecutive probability for a previous digit and a consecutive probability for a subsequent digit. 10. A sequence according to claim 8, characterized in that the two-number consecutive probability is represented by binary values of high and low for the probability that a number follows one number and the probability that that number follows the previous number. Number recognition device. 11. Claims 7 and 8 characterized in that a digit sequence probability changing means (17) is provided for changing the digit sequence probability stored in the digit sequence probability storage section (16) in accordance with the writing order of the input digits. Or the continuous number recognition device described in 9.