JPH0441402Y2 - - Google Patents

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a handwritten character recognition device used as an input device for, for example, a word processor or a microcomputer system.

[Prior art]

As a Japanese input device for a microcomputer system and a word processor using the same, in addition to key input using a keyboard, it detects characters handwritten on a tablet with a character input pen and the trajectory of the character input pen. A so-called handwritten character recognition device is used.

By the way, in conventional handwritten character recognition devices, when characters are handwritten with a character input pen on the tablet surface, which is the character input surface, the locus (stroke) of the character input pen corresponding to each stroke of the character is converted into time series coordinate data. This data is used to recognize handwritten characters. In addition, in order to recognize each character, the end of inputting each character is detected, and the stroke string input from the end of inputting each character until the end of inputting the next character is detected. Series coordinate data must be recognized as one character's worth of data.

Conventionally, the end of inputting a single character has been detected by, for example, touching a character input pen to a predetermined area on the tablet, or, if input frames for each character are preset on the tablet, inputting a character into the next input frame. A common method is to input characters, or even to interrupt character input using a character input pen on a tablet for a certain period of time or more.

[Problem that the invention attempts to solve]

By the way, in the conventional handwritten character recognition device as described above, for example, if the operator interrupts the character input operation in the middle of inputting one character due to some reason, or if the operator , for example, if you forget to input a character such as bia, 旁, dot, etc., as mentioned above, enough time has elapsed for it to be detected as the end of inputting one character.
Alternatively, a situation may arise in which the next character is input and the character in the middle of input is erroneously recognized as a character different from the operator's intention. In order to correct a character that has been erroneously recognized in this way, it is necessary for the operator to once cancel the erroneously recognized character and then input the character that the operator originally intended to input by hand again. In this case, if the operator has to re-enter characters with a small number of strokes, such as ``earth'' or ``人'', the burden on the operator is not so great; Re-inputting characters by hand is a burden on the operator, and also reduces overall processing efficiency.

The present invention was developed in view of the above-mentioned problems, and is intended to replace characters that are incorrectly recognized as having been input when halfway through a character has been input, with missing characters that should originally have been input subsequently. To provide a handwritten character recognition device which can be easily and quickly corrected by simply adding additional information by handwriting input.

[Means for solving problems]

The handwritten character recognition device of the present invention detects a part of a newly input character, that is, a stroke, which is added to an already input character, according to predetermined conditions, and the detected stroke and this stroke are added to the newly input character. By configuring the system to re-recognize the time-series coordinate data of the stroke string of a character that has already been inputted as one character's worth of data, characters that are misrecognized due to interruptions in input can be easily removed. It is possible to modify it.

The present invention provides a handwritten character recognition device that detects and recognizes each stroke of a character handwritten by a character input pen on a character input surface of a character input device as time-series coordinate data of the trajectory of the character input pen on the input surface. , single-character input end detection means for detecting the end of handwritten input of one character on the character input surface according to a first predetermined condition; and detection of the end of input of each single character and the next end detection by the single-character input end detection means. A memory that stores the time-series coordinate data of a stroke string input during the process as time-series coordinate data of each character, and a memory that stores characters input by hand based on the time-series coordinate data of one character stored in the memory. a second predetermined character recognition unit that recognizes that the newly input stroke is a continuation stroke of one of the character stroke strings already stored in the memory as time series coordinate data for one character; Continuation stroke detection means detects according to conditions; and a stroke detected by the continuation stroke detection means as a continuation stroke of any stroke string already stored in the memory as time-series coordinate data for one character. and control means for storing the time-series coordinate data in the memory together with the time-series coordinate data of the stroke string for one character as the time-series coordinate data for one character, and causing the character recognition unit to recognize this again. It is characterized by

[Effect]

With the handwritten character recognition device of the present invention, a character that is erroneously recognized as another character before the input is completely completed can be easily corrected by adding the part that was not input and inputting it by hand.

〔Example〕

Hereinafter, the present invention will be described in detail based on drawings showing embodiments thereof.

FIG. 1 is a block diagram showing a schematic configuration of a handwritten character recognition device according to the present invention (hereinafter referred to as the device of the present invention).

1 in the figure is a handwritten character input device; for example, when handwritten characters are input with a character input pen on a character input surface on which an X-Y coordinate system is set, the trajectory of the character input pen corresponding to one stroke of the character Detected as time series coordinate data in the X-Y coordinate system,
Detect each stroke of handwritten characters. The time-series coordinate data of each stroke read by this handwritten character input device 1 is given to a time-series coordinate data holding recording device (hereinafter referred to as memory) 2.

The memory 2 includes a plurality of time-series coordinate data holding areas 21 that store time-series coordinate data on a character-by-character basis;
22...2j. The time series coordinate data stored in the time series coordinate data holding areas 21, 22, . . . is given to the handwritten character recognition unit 3 as data for one character.

The handwritten character recognition unit 3 specifies and recognizes the time-series coordinate data given from the memory 2, for example, by referring to a recognition dictionary (not shown).

4 in the figure is a one-character input end detection means, which allows the operator to complete the handwritten input of one character;
Alternatively, the end of the handwritten input of one character is detected by not touching the character input surface of the handwritten character input device 1 with the character input pen for a certain period of time or more. Note that operations by the operator to end handwritten input of a single character include, for example, touching a character input pen to a predetermined part on the character input surface to instruct the end of character input, or touching a character entry frame preset on the character input surface. It is conceivable that the character input pen touches the next frame.

Reference numeral 5 in the figure indicates a continued stroke detection means, in which a stroke string newly inputted on the character input surface is already detected as one character by the above-mentioned single character input end detection means 4, and its time series coordinate data is stored in the memory. It is detected that the stroke is a continuation of the character stroke string stored in any of the time-series coordinate data holding areas 21, 22, .

Specifically, a newly input handwritten stroke is a stroke that was input immediately before it (therefore, the character that includes this stroke is the character that was input at the closest point in time, and the stroke string of that character is If the time-series coordinate data is input in the same character entry frame as the time-series coordinate data holding area 21, 22, etc. at the end of the memory 2 where the time-series coordinate data is stored, the new handwritten The time-series coordinate data of the input stroke is stored as part of the last input character in one of the time-series coordinate data holding areas 21, 22, which stores the time-series coordinate data of the stroke string of that character.
is stored as one character's worth of data. Further, when a newly inputted stroke is already inputted by the coordinate values of the time-series coordinate data, the time-series coordinate data holding areas 21, 22 of the memory 2...
Detected as part of a character stored as time-series coordinate data of a stroke string for one character. In this case, the time-series coordinate data of the newly inputted stroke is stored in one of the time-series coordinate data storage areas that stores the time-series coordinate data of the stroke string of the character in which the stroke was detected as part of the time-series coordinate data. 21, 22...

6 in the figure is a control device (hereinafter referred to as a memory control device) of the recording device 2 for holding time-series coordinate data, and the above-mentioned continued stroke detection means 5 detects a newly inputted stroke as part of an already inputted character. When the stroke is detected by the corresponding time series coordinate data holding area 21, 22...
At the same time, the time-series coordinate data stored in either of the time-series coordinate data holding areas 21, 22, etc. (time-series coordinate data of newly input strokes and time-series coordinate data of already input strokes) (time-series coordinate data combined with time-series coordinate data) is given to the handwritten character recognition unit 3 for re-recognition.

The operation of the present device configured as above will be explained below. Note that FIG. 2 is an explanatory diagram of the operation of the present device, and a case where "identification example" is input will be described.

First, as shown in FIG. 2a, it is assumed that the first character ``Kiwa'' is handwritten into the first character entry frame 71. The time-series coordinate data of the stroke string of the first handwritten "word" is stored in the time-series coordinate data holding area 21 at the beginning of the memory 2. For example, as shown in Figure 2b, if only the first character entry frame 71 is filled with the word ``sense'', then the second character ``another'' is entered in the next character entry frame 72. When "〓" is partially input, the one-character input end detection means 4 detects the end of character input into the first character entry frame 71. As a result, the memory control device 6 provides the time-series coordinate data stored in the top time-series coordinate data holding area 21 of the memory 2 to the handwritten character recognition unit 3.
The handwritten character recognition unit 3 recognizes the character handwritten and input into the first character entry frame 71 as "word".

After this, as shown in Figure 2c, when the character ``〓'' is added to the first character entry frame 71,
Continuation stroke detection means 5 detects the first character entry frame 7
1 is detected as a continuation of the stroke string that has already been input into the first character entry frame 71 and recognized by the handwritten character recognition unit 3 as the character ``word''. As a result, the memory control device 6 stores the time-series coordinate data of each stroke of the stroke string "〓" added in the character entry frame 71 at the beginning in the time-series coordinate data holding area 21 at the beginning of the memory 2. It is added to the time-series coordinate data of the stroke string of "word" and stored sequentially.

In this way, after the handwriting input of the character "KI" into the first character entry frame 71 is completed, as shown in FIG. When "" is input as part of the uninputted part of "", the single character input end detection means 4
detects that the handwritten input of one character into the first character entry frame 71 has been completed. As a result, the memory control device 6 transfers the time-series coordinate data (data of the handwritten character "KI") currently stored in the time-series coordinate data holding area 21 at the beginning of the memory 2 to the handwritten character recognition unit 3. Let them give it to them and let them re-understand it.

Then, as shown in FIG. 2e, when the remaining part "〓" of the character "betsu" is input into the second character entry frame 72, the stroke detection means 5 continues to detect the new input in the same manner as described above. The stroke string “〓” is detected as the remaining strokes of “〓” that has already been input into the second character entry frame 72, and its time series coordinate data is sequentially stored in the second time series coordinate data holding area 22 of the memory 2. Make me remember.

Furthermore, as shown in FIG. The end of character input to 72 is detected. As a result, the memory control device 6 provides the time-series coordinate data (data for the handwritten character "different") stored in the second time-series coordinate data holding area 22 of the memory 2 to the handwritten character recognition unit 3. Let them know again.

Lastly, as shown in FIG. It is detected that this is additional writing to the character entry frame 73. As a result, the memory control device 6
The time-series coordinate data of each stroke handwritten in the third character entry frame 73 is stored in the third time-series coordinate data holding area 23 of the memory 2, and the time-series coordinate data of the stroke "'" that is already stored is stored in the third time-series coordinate data holding area 23 of the memory 2. Memorize them sequentially. Then, when the last stroke of the character "example" is handwritten input into the third character entry frame 73 and a predetermined period of time has elapsed without the character input pen touching the character input surface, the one character input end detection means 4 detects The end of character input into the character entry frame 73 of No. 3 is detected. Thereby, the memory control device 6 gives the time-series coordinate data stored in the third time-series coordinate data holding area 23 of the memory 2 to the handwritten character recognition unit 3 for recognition.

〔effect〕

As described above, in the proposed device, even if character input is interrupted midway and character recognition is performed only on the part input up to that point, resulting in erroneous recognition, the remaining part can be added. This allows for easy and quick correction. Therefore, it is no longer necessary to cancel the erroneously recognized character and input the entire character by hand again as in the conventional method, which reduces the burden on the operator and improves the processing efficiency of character input as a whole.

[Brief explanation of the drawing]

The drawings show an embodiment of the present invention.
The figure is a block diagram showing the general structure of the present device, and FIG. 2 is a schematic diagram for explaining its operation. 1... Handwritten character input device, 2... Recording device (memory) for holding time-series coordinate data, 3... Handwritten character recognition section, 4... Single character input end detection means, 5
...Continued stroke detection means, 6...Memory control device.

Claims (1)

[Claims for Utility Model Registration] 1. Detecting and recognizing each stroke of a character handwritten by a character input pen onto a character input surface of a character input device as time-series coordinate data of the trajectory of the character input pen on the input surface. The handwritten character recognition device includes: a single character input end detection means for detecting the end of handwritten input of one character on the character input surface according to a first predetermined condition; and detection of end of input of each single character by the single character input end detection means. A memory that stores the time-series coordinate data of the stroke string input between the next end detection as time-series coordinate data for each character, and handwriting based on the time-series coordinate data for one character stored in the memory. A character recognition unit that recognizes an input character, and a character recognition unit that recognizes that a newly input stroke is a continuation of a stroke string of one of the characters already stored in the memory as time series coordinate data for one character. a continuation stroke detection means that detects according to a second predetermined condition; Control means for storing time-series coordinate data of the detected stroke in the memory together with time-series coordinate data of a stroke string for one character as time-series coordinate data for one character, and causing the character recognition unit to recognize this again. A handwritten character recognition device comprising: 2. The handwritten character recognition device according to claim 1, wherein the first predetermined condition is a contact of a predetermined portion of the input surface with the character input pen. 3. The handwritten character recognition device according to claim 1, wherein the first predetermined condition is that the character input pen remains in a non-contact state with the input surface for a predetermined period of time. 4. The first predetermined condition is that the character input on the input surface using the character input pen is performed in the input frame next to the character input frame preset on the input surface. A handwritten character recognition device according to claim 1. 5. Handwritten character recognition according to claim 1, wherein the second predetermined condition is that a newly input character string is input within a predetermined distance from an already input character string. Device. 6. A utility model in which the second predetermined condition is that a newly input stroke is input into an input frame in which a character string has already been input, of character input frames preset on the input surface. A handwritten character recognition device according to claim 1.