JPH0863552A - Device for discriminating capital or small letter in handwritten character string - Google Patents

Abstract

PURPOSE: To discriminate a capital letter or a small letter in character input without a character frame. CONSTITUTION: A character string segmenting means 2 segments a character string from a stroke code string inputted from a table 1. A recognition means 3 recognizes the character based on stroke information from a starting stroke number to a finishing stroke number. A capital or small letter discriminating means 40 detects whether the character recognized by the recognition means 3 is the same-shaped and duplicated character or not. A threshold value deciding means 42 obtains the threshold of the same-shaped and duplicated character detected by the capital or small letter discriminating means 40. A character size comparing and deciding means 41 obtains the ratio of the featured value of the detected character to the featured value of a capital letter other than the detected character, then it is discriminated that the detected character is small letter when the ratio is smaller than the threshold value and is capital letter when the ratio is larger than the threshold value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a handwritten character string uppercase / lowercase determination device for determining the uppercase / lowercase of a handwritten character string.

[0002]

2. Description of the Related Art Conventionally, techniques in such a field include:
For example, some documents were described in the following documents. Reference: Japanese Examined Patent Publication No. 1-50954 Conventionally, in a handwritten kanji input device that writes characters one by one in a character frame, when a character is handwritten on an input tablet, character recognition processing is executed for each character, and The recognition result is output to a display device or the like. By the way, some of the input characters are the same characters such as kana characters “ya” and “ya”, “yu” and “yu”, “yo” and “yo”, and “tsu” and “tsu”. There are some cases in which there is a distinction between uppercase and lowercase letters, and if there is a character input of this kind, it is necessary for the device side to clearly distinguish the two. As a method for solving this, a method has been devised as described in the above-mentioned document, in which uppercase and lowercase letters are determined according to the size of an input character with respect to the size of a character frame.

[0003]

However, the conventional handwritten character string case determination device has the following problems. Good man by handwritten kanji input
As a method of realizing a machine interface, it is desired to input a character string written on a blank sheet without a character frame so as not to burden the writer. Only valid if present.
In other words, the size of the characters entered in a freely written character string without a character frame varies considerably if the writer changes, the time and place of the same writer, and even the line changes on the same screen. Since it occurs, the size of the character frame cannot be uniquely determined, and it is impossible to distinguish between uppercase and lowercase letters.

[0004]

In order to solve the above-mentioned problems, a first aspect of the present invention provides a character cutting-out means for cutting out a character from a coordinate data string obtained from a coordinate input device, and one character by the character cutting-out means. A handwritten character string upper / lower case determination device provided with a recognition means for recognizing a character from a cut out coordinate data string is provided with the following means. That is, the possibility that the character recognized by the recognizing means is lowercase is judged, and if there is the possibility that the character is lowercase, it is already judged whether the size of the character and whether the character is recognized by the recognizing means is uppercase or lowercase. By comparing the size of the other character based on the coordinate data string cut out by the character cutting means, there is provided an upper / lower case determining means for determining whether the character is lowercase or uppercase. . In the second invention, the upper / lower case determining means of the first invention comprises the following means. That is, among the characters recognized by the recognizing means, an uppercase / lowercase detecting means for detecting a target character having the same or similar uppercase and lowercase fonts, and whether the target character detected by the uppercase / lowercase detecting means is an uppercase character. Threshold value determining means for determining a threshold value for determining whether the character is lowercase, and the size of the target character detected by the uppercase / lowercase detecting means based on the coordinate data string cut out by the character cutting means. The ratio between the feature amount and the feature amount which is recognized by the recognition means and which has already been determined to be uppercase or lowercase and which represents the size of another character is obtained, and this ratio is determined by the threshold value determination means. And a character size comparison / determination means for comparing the threshold values of the characters and determining whether the target character is uppercase or lowercase. In a third invention, the threshold value determining means of the second invention determines a threshold value for each target character. Fourth
In the second or third aspect of the invention, the input rule determination for determining whether or not the character recognized by the recognition means and the character immediately preceding the character satisfy the necessary condition that the character is lowercase. Means are provided.

[0005]

According to the first aspect of the invention, since the handwritten character string upper / lower case determining device is configured as described above, the upper / lower case determining means can make the character lowercase among the characters recognized by the recognizing means. Judge sex. Then, if there is a possibility that it will be a lowercase letter, the character and another uppercase letter recognized by the recognizing means are compared based on the coordinate data string cut out by the character slicing means, and the size is compared. , Or upper case. According to the second invention, the uppercase / lowercase detection means is
Among the characters recognized by the recognition means, a target character having the same or similar uppercase and lowercase fonts is detected. The threshold value determining means determines a threshold value for determining whether the target character detected by the uppercase / lowercase detecting means is lowercase or uppercase. Character size comparison and determination means
The ratio between the feature amount representing the size of the target character detected by the uppercase / lowercase detection unit and the feature amount representing the size of the character recognized by the recognition unit and already determined to be uppercase or lowercase, is obtained, This ratio is compared with the threshold value of the character determined by the threshold value determining means to determine whether the character is uppercase or lowercase. According to the fourth aspect of the invention, the input rule determination means determines whether or not the required condition that the character is a lower case is satisfied by the combination of the character recognized by the recognition means and the character immediately preceding the character.
Therefore, the above problem can be solved.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First Embodiment FIG. 1 is a block diagram of a handwritten character string upper / lower case determination device showing a first embodiment of the present invention. This handwritten character string upper / lower case determination device has a tablet 1 for inputting handwritten characters. The character string cutout unit 2 is connected to the output side of the tablet 1, and the recognition unit 3 and the upper / lower case determination unit 4 are connected to the output side of the character string cutout unit 2. The case determination means 4 is
The case detection means 40 connected to the output side of the recognition means 3 and the character size comparison determination means 41 and the threshold value determination means 4 connected to the output side of the case detection means 40.
2 and. The character size comparison and determination means 41 is
It is also connected to the output side of the character string cutout means 2 and the threshold value determination means 42. The tablet 1, which is the coordinate input device, inputs a character string by handwriting, and the coordinate value of the writing point is fixed at a fixed time interval for each stroke from when the pen tip contacts the tablet 1 until the pen tip leaves the tablet 1. Is taken in and sent to the character string cutout means 2.
The character string cutout unit 2 is a unit that cuts out a character from the coordinate value of the writing point of each stroke and sends the result to the recognition unit 3 and the uppercase / lowercase determination unit 4. Recognition means 3
Is a means for recognizing the characters cut out by the character string cutout means 2.

Among the characters recognized by the recognition means 3, the upper / lower case determination means 4 includes the characters "ya" and "ya", "yu" and "yu", "yo" and "yo", and "tsu". It is a means for determining uppercase and lowercase letters such as "and". The upper-case and lower-case detection means 40 has a character recognized in the recognition result by the recognition means 3 such as "ya", "yu", "yo", and "tsu" in the same upper and lower case letters (hereinafter referred to as an isomorphic duplicate character). Alternatively, a similar target character is detected. The threshold value determining means 41 is means for determining a threshold value for determining whether the target character detected by the uppercase / lowercase detecting means 40 is uppercase or lowercase. The character size comparison / determination means 42 uses the feature amount representing the size of the target character detected by the upper / lower case detection means 40 from the coordinate value of the character cut out by the character string cutout means 2 and whether it is uppercase or lowercase. A feature amount representing the size of another character, which is already determined, is obtained, the ratio of the feature amounts is obtained, and this ratio is compared with the threshold value for the target character determined by the threshold value determining means 42. However, it is a means for determining whether the target character is uppercase or lowercase.

Next, the operation of the handwritten character string upper / lower case determination device of FIG. 1 will be described. First, the stroke code string input from the tablet 1 is a character string cutout unit 2
Sent to FIG. 2 is a diagram showing an example of input characters, FIG. 3 is a diagram showing a result (image) cut out by the character string cutout unit 2, and FIG. 4 is cut out by the character string cutout unit 2. It is a figure which shows the result (feature) which was recognized, and FIG. 5 is a figure which shows the result recognized by the recognition means 3. For example, if a character string as shown in FIG. 2 is input, the character string cutout unit 2 cuts out the character as shown in FIG. 3, and as shown in FIG. 4, the start stroke number and end of the character for each stroke. The stroke number, the minimum value of the X coordinate, the maximum value of the X coordinate, the minimum value of the Y coordinate, and the maximum value of the Y coordinate that represent the circumscribed rectangle cut out as a character are output. The recognition means 3 recognizes the stroke information from the start stroke number to the end stroke number, and
The recognition result like is output.

FIG. 6 shows the uppercase child character determination means 4 in FIG.
5 is a flowchart showing the processing contents of FIG. In step S1, the uppercase child character determination means 40 in FIG. 1 first extracts one character from the recognized result, and proceeds to step S2. In step S2, it is detected whether the extracted character is an isomorphic duplicate character, and if it is an isomorphic duplicate character, the process proceeds to step S3. In the case of FIG. 5, the second "yu", the third "u", the fifth "tsu", the ninth "yo", the tenth "tsu", and the thirteenth "tsu" have the same shape duplication. It is detected as a character. No processing is performed on the other characters, so the process proceeds to step S8. Step S3
In, the character size comparison and determination means 41 obtains the corresponding feature amount from the character cutout result. In step S4, the threshold value determining means 42 detects the uppercase / lowercase detection means 40.
To obtain the threshold value of the isomorphic duplicate characters detected by. Figure 7
FIG. 4 is a diagram showing a threshold value for each character, in which an upper column is an uppercase target character and a lower column is a lowercase target character. In Fig. 7, there are similar combinations of uppercase and lowercase fonts, such as "la" and "u", and "hu" and "tsu", but this is small despite being written as "u". This is because, in the case where it has been recognized as "La", the uppercase / lowercase detection means 40 cannot detect it unless only the combination of "u" and "u" is recognized.

FIG. 8 shows the value TH obtained by the threshold value determining means.
FIG. When the detected character (hereinafter referred to as the detected character) is the nth character, the coordinate value (Xmin _n , Xmax of the detected character is obtained from the result output by the character slicing means 2.
_n , Ymax _n , Ymin _n ) is obtained, and the feature amount L _n shown in the following equation (1) is calculated. If _{L n = (Ymax n -Ymin n} ) ··· (1) immediately before the detection character is not a lower case (second in FIG. 5,
The fifth, ninth, and thirteenth) calculate the feature amount L _n-1 immediately before the detected character, determine whether or not the condition L _n / L _n-1 <TH for small letters is satisfied, and satisfy this condition. If so, it is determined to be lowercase in step S6 and the process proceeds to step S8. If this condition is not satisfied, it is determined to be uppercase in step S7 and the process proceeds to step S8.

If the character just before the detected character is a small letter,
The upper-case feature amount L _nm immediately before the detected lower-case letters are removed, and the lower-case condition L _n / L _nm <TH
Is satisfied. If this condition is satisfied, it is determined to be a lowercase character in step S6 and the process proceeds to step S8. If this condition is not satisfied, it is determined to be an uppercase character in step S7 and the process proceeds to step S8. If the detected character is the first character, the characteristic amount L _{n + m} of the upper case immediately after the detected character is calculated, and the condition of the lower case L _n / L _{n + m} <TH
Is satisfied. If this condition is satisfied, it is determined to be a lowercase character in step S6 and the process proceeds to step S8. If this condition is not satisfied, it is determined to be an uppercase character in step S7 and the process proceeds to step S8. In step S8, the above step S1 is performed for all the characters of the recognition result.
~ It is determined whether the process up to step S7 has been performed. If all the characters have been performed, the process returns to step S1, the above process is repeated, and if all the characters have been performed, the process ends. FIG. 9 is a diagram showing the upper / lower case determination result.

As described above, the first embodiment has the following advantages. In a character string written on a blank sheet without a character frame, "ya" and "ya", "yu" and "yu", "yo" and "yo", "tsu" and "tsu", etc. Even if a character exists, it is possible to eliminate the influence of the character size outside the range by comparing the size of the character with the narrow range immediately before the character and comparing the size, and by setting a threshold for each character. However, it is possible to effectively input Japanese because the letters of interest can be case sensitive. For example, when the second character "Yu" is detected, the circumscribed rectangle coordinate value immediately before the detected character is (Xmin ₁ , Xmax ₁ , Ymin ₁ , Ymax ₁ ) = (0,
100, 20, 150), and the circumscribed rectangle coordinate value of the detected character is (Xmin ₂ , Xmax ₂ , Ymin ₂ , Ymax ₂ ) = (11
4,178,10,100), and L ₁ = (100-0) × (150-20) = 13000 L ₂ = (178-114) × (100-10) = 567
We obtain 0 L ₂ / L ₁ = 0.44 <TH (0.6). Therefore, it is determined to be a small letter "yu".

Second Embodiment FIG. 10 is a block diagram of a handwritten character string upper / lower case determining apparatus according to a second embodiment of the present invention, in which elements similar to those in FIG. It is attached. The second embodiment is different from the first embodiment in that the character recognized by the recognition means 3 and the character immediately before this character are provided on the input side of the case detection means 40 in the case determination means 4. That is, the input rule determination means 43 is provided for determining whether or not the character satisfies the required condition that the character is a lowercase character by combining the characters. As shown in FIG.
Is provided on the output side of the character cutting means 2 and the recognition means 3. The upper case / lower case determining means 4 is provided with an input rule determining means 43. Input rule determination means 43
An uppercase / lowercase detection means 40 is connected to the output side of the. The other configuration is the same as that of FIG. The input rule determination means 43 is a means for determining whether or not the detected character and the character immediately before it match the input rule (required condition that the detected character is lowercase).

The operation of the handwritten character string upper / lower case determination device of FIG. 10 will be described below. First, the stroke code string input from the tablet 1 is used by the character string slicing device 2.
Sent to 11 is a diagram showing an example of an input character, FIG. 12 is a diagram showing a result (image) cut out by the character string cutout unit 2, and FIG. 13 is a diagram showing a result recognized by the recognition unit 3. FIG. For example, FIG.
If a character string such as 1 is input, the character string cutout unit 2 cuts out the character as shown in FIG. 12, and the start stroke number, end stroke number, and character of the character are cut as in the first embodiment. The minimum value of the X coordinate, the maximum value of the X coordinate, the minimum value of the Y coordinate, and the maximum value of the Y coordinate that represent the circumscribed rectangle cut out as are output. The recognition means 3 recognizes the stroke information from the start stroke number to the end stroke number and outputs the recognition result as shown in FIG. FIG. 14 is a schematic flowchart showing the operation of the upper case child character determination means 4 in FIG. FIG.
FIG. 6 is a diagram showing an example of an input rule. FIG. 16 is a diagram showing an input rule determination result and a final result. Hereinafter, the operation of the upper case child character determination means 4 will be described with reference to these figures.

In the uppercase child character determination means 4 in FIG.
In step S11, one character is first extracted from the recognized result, and the process proceeds to step S12. In step S12, it is determined whether or not the extracted character is the first character, and if it is the first character, the character is not an isomorphic duplicate character and the process returns to step S11. If so, the process proceeds to step S13. Step S13
In step S14, it is detected whether the extracted character is an isomorphic duplicate character.
Proceed to. In the case of FIG. 13, the third “ya” and the seventh “ya” are detected as isomorphic duplicate characters. No processing is performed on the other characters, and the process returns to step S11. In step S14, when the isomorphic duplicate character is detected in step S13, one character immediately before the detected isomorphic duplicate character is extracted. In the case of FIG. 13, the second “letter” is extracted for the third “ya” and the sixth “chi” is extracted for the seventh “ya”. Step S15
In step S16, it is determined whether the combination of these preceding and following characters is a character existing in the input rule shown in FIG. If it is a nonexistent character, the isomorphic duplicate character is not a lower case character, and therefore the process proceeds to step S20. In the case of FIG. 13, since “letter” and “ya” do not satisfy the input rule, it is determined that “ya” is uppercase, and “chi” and “ya” satisfy the input rule.
It is determined whether "ya" is uppercase or lowercase.

In step S16, the character size comparison / determination means 41 obtains the corresponding feature amount from the result of character extraction. In step S17, the threshold value determining means 4
In 2, the threshold value of the isomorphic duplicate characters detected by the uppercase / lowercase detecting means 40 is obtained. Step S18, Step S
19, in step S20, step S in FIG.
5, the same processing as step S6 and step S7 is performed.
In step S21, it is determined whether or not the processes of steps S11 to S20 have been performed for all the characters of the recognition result. If all the characters have been processed, the process returns to step S11 and the above process is repeated to perform all the characters. If not, the process ends. FIG.
Shows the input rule judgment result and the final result. As described above, the second embodiment has the following advantages. In addition to the same advantages as those of the first embodiment, in the first embodiment, the uppercase child character detection means 4 detects all characters that may be uppercase letters and lowercase letters, and detects uppercase letters or lowercase letters. It was determined whether there is any, but the input rule determination means 4
By providing 3, it is possible to eliminate a state that is almost impossible when a general word or sentence is input, and it is possible to input Japanese effectively.

Third Embodiment FIG. 17 is a block diagram of a handwritten character string upper / lower case determination device according to a third embodiment of the present invention, in which elements similar to those in FIG. It is attached. The third embodiment differs from the second embodiment in that the input rule determining means 43 is
Switch means 44 between the upper case and lower case detection means 40
Is provided. The switch means 44 determines whether or not the writer uses the input rule determination means 43. Next, the operation of the handwritten character string upper / lower case determination device of FIG. 17 will be described. The writer decides whether or not to use the input rule determining means 43. When using the input rule determining means 43, the input rule determining means 43 and the uppercase / lowercase detecting means 40 are connected by the stitching means 44 and input. When the rule determination means 43 is not used,
The input rule determination means 43 and the upper / lower case detection means 40 are separated by the stitch means 44. As described above, in the third embodiment, it is possible to determine whether or not to use the input rule determining means 43, so that it is possible to determine the uppercase and lowercase letters that are more elastic.

The present invention is not limited to the above embodiment, and various modifications can be made. The following are examples of such modifications. (A) In the present embodiment, uppercase child characters are determined using the height ratio of the circumscribing rectangle, but other methods such as the ratio of the diagonal distance of the circumscribing rectangle, the ratio of the lateral distance, and the area You may use a ratio, a gravity center, etc. (B) In this embodiment, the threshold value for each character is fixed and used, but it is also possible to set the threshold value more finely with a configuration that can be freely set by the writer. (C) In the present embodiment, the example applied to the online handwriting device has been described, but it is also applicable to the OCR device.

[0019]

As described above in detail, the first to the fourth
According to the present invention, since the upper / lower case determination means is provided, even if there is a lowercase character in a character string written on a blank sheet without a character frame, the uppercase and lowercase letters can be distinguished, so that it is effective in Japan. You can enter words.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a handwritten character string case recognizing device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of an input character string.

FIG. 3 is a diagram showing a result (image) cut out by a character string cutout unit.

FIG. 4 is a diagram showing a result (feature amount) cut out by a character string cutout unit.

FIG. 5 is a diagram showing a result recognized by a recognition means.

FIG. 6 is a flowchart showing the processing contents of upper / lower case determination means in FIG.

FIG. 7 is a diagram showing a threshold value for each character.

FIG. 8 is a diagram showing values obtained by threshold value determining means.

FIG. 9 is a diagram showing a case determination result.

FIG. 10 is a configuration diagram of a handwritten character string case recognizing device according to a second embodiment of the present invention.

FIG. 11 is a diagram showing an example of an input character string.

FIG. 12 is a diagram showing a result (image) cut out by a character string cutout unit.

FIG. 13 is a diagram showing a result recognized by a recognition means.

FIG. 14 is a flowchart showing the processing contents of upper / lower case determination means in FIG.

FIG. 15 is a diagram showing an example of an input rule.

FIG. 16 is a diagram showing an input rule determination result and a final result.

FIG. 17 is a configuration diagram of a handwritten character string case recognizing device according to a third embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 tablet 2 character string cutout means 3 recognition means 4 upper / lower case determination means 40 upper / lower case detection means 41 character size comparison determination means 42 threshold value determination means 43 input rule determination means 44 switch means

Claims (4)

[Claims] 1. A character cutting means for cutting a character from a coordinate data string obtained from a coordinate input device, and a recognition means for recognizing a character from a coordinate data string cut out as one character by the character cutting means. In the handwritten character string case determination device, the possibility that the character recognized by the recognition means becomes a lowercase character is determined, and if there is a possibility that it becomes a lowercase character, the size of the character and whether the character is recognized by the recognition means is an uppercase character. By comparing the sizes of other characters, which have already been determined to be lowercase, based on the coordinate data string cut out by the character cutting means, it is determined whether the character is lowercase or uppercase. An upper / lower case determination device for handwritten character strings, which is provided with a lower case determination means. 2. The uppercase / lowercase determination means detects uppercase / lowercase detection means for detecting target characters having the same or similar uppercase and lowercase fonts among the characters recognized by the recognition means, and the uppercase / lowercase detection means. A threshold value determining means for determining a threshold value for determining whether the detected target character is uppercase or lowercase, and based on the coordinate data string cut out by the character cutout means, by the uppercase / lowercase detection means. The ratio of the feature amount representing the size of the detected target character and the feature amount representing the size of another character, which is recognized by the recognition means and has already been determined to be uppercase or lowercase, is obtained, and this ratio Comparing the threshold values of the characters determined by the threshold value determining means, the character size for determining whether the target character is uppercase or lowercase. A handwritten character string upper / lower case determination device, characterized in that it comprises a size comparison determination means. 3. The handwritten character string uppercase / lowercase determination means according to claim 2, wherein the threshold value determination means determines a threshold value for each target character. 4. An input rule determining means is provided for determining whether or not the character recognized by the recognizing means and a character immediately preceding the character meet a requirement that the character is lowercase. The handwritten character string upper / lower case determination device according to claim 2 or 3.