JPH1097591A - Frameless on-line character recognition device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily detect the line direction of a character string and decide whether the character string is written longitudinally or laterally without previously fixing a way of writing the character string and the line direction of the character string. SOLUTION: A handwriting inputted on a coordinate input surface is detected by a handwriting input part 11 as a time series of two-dimensional coordinates and stored in a storage part 12. A base line detection part 13 detects a base line from strokes constituting the handwriting (indicated by the coordinate series) stored in the storage part 12. A character string information extraction part 14 detects the line direction of the character string in the handwriting from the base line and decides whether the character string is written longitudinally or laterally on the basis of the position relation between the detected base line and the character string in the handwriting according to the restrictions of the predetermined position relation between the character string and base line. A line correction part 15 rotates the character string in the handwriting according to the line direction detection result and longitudinal/lateral writing decision result of the character string to enable a recognizing process by a character recognition part 16.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a frameless online character recognition device suitable for recognizing, in real time, a character string of handwriting handwritten in an area without a frame, and more particularly to a line direction of a character string, and more particularly, to a character string line direction. The present invention relates to an online character recognition device having a function of detecting vertical / horizontal writing of a character string.

[0002]

2. Description of the Related Art In recent years, an online character recognition device that recognizes a character string written on a coordinate input surface of a tablet or the like in real time has attracted attention. One such type of online character recognition device is a frameless online character recognition device that recognizes handwritten characters written in a frameless region.

[0003] One of the following two methods is generally applied to the conventional frameless online character recognition device. The first method restricts the writing direction,
If the character string is written horizontally, the writing direction is limited to the horizontal direction (x-axis direction) shown in FIG. 19, and if it is written vertically, the writing direction is limited to the vertical direction (y-axis direction) shown in FIG.

In the frameless online character recognition apparatus to which the first method is applied, as a result, it is necessary to determine whether the text is written vertically or horizontally and to detect the line direction of the character string (the direction of the character line). Character recognition can be performed without performing.
However, the first method imposes restrictions on writing, and thus imposes a burden on the user (writer).

On the other hand, the second technique is to recognize a character string written in an arbitrary direction in an arbitrary writing method (vertical writing or horizontal writing) without limiting the writing direction. In the frameless online character recognition device to which the second method is applied, the degree of freedom of the user is increased. However, unlike the first method, before the recognition, the character is written vertically or horizontally. Must be determined and the line direction of the character string must be detected. For this reason, processing takes time, and recognition accuracy deteriorates when detection fails.

[0006]

As described above, in the conventional frameless on-line character recognition technology, it is determined whether the text is written vertically or horizontally and the line direction of the character string (the direction of the character line). If the detection is not required, the writing direction must be restricted, which imposes a burden on the user. Conversely, in order not to restrict the writing direction, it is necessary to write vertically before recognition. Or horizontal writing and the direction of the character line must be detected, which causes a problem that the recognition speed and the recognition accuracy are reduced.

The present invention has been made in consideration of the above circumstances, and has as its object to fix a character string in an arbitrary writing method and in an arbitrary direction without fixing in advance the method of writing a character string and the line direction of the character string. , The line direction of a character string necessary for recognition of the character string can be detected, and the vertical / horizontal writing of the character string can be easily determined, thereby improving the recognition speed and the recognition accuracy. An object of the present invention is to provide a frameless online character recognition device.

[0008]

A frameless online character recognition apparatus according to a first aspect of the present invention includes a handwriting input means for detecting and inputting a handwriting written on a coordinate input surface as a time series of two-dimensional coordinates. A plurality of strokes constituting the handwriting input by the handwriting input means, a baseline detection means for detecting a stroke indicating the inclination of the character string in the handwriting, and a baseline detection means for detecting the baseline stroke. Character string information extracting means for detecting the line direction of the character string from the obtained baseline stroke, and rotating the character string in the handwriting based on the detection result of the character string line direction by the character string information extracting means. It is characterized by comprising a line correcting means and a character string recognizing means for recognizing a character string rotated by the line correcting means.

In such a configuration, the user only writes a stroke indicating the inclination of the character string at the time of writing the character string, while allowing the user to write and input the character string in an arbitrary manner and in an arbitrary direction. Thus, the line direction of the character string in the handwriting can be easily and automatically detected based on the stroke (base line). Thereby, the recognition speed and the recognition accuracy can be improved. However, since it is not easy to detect whether the character string is written vertically or horizontally, the writing method and the line direction of the character string are lower than those fixed in advance. Therefore, by restricting how to write a character string (vertical writing or horizontal writing), that is, by only allowing a character string to be written and input in any direction,
It is also possible to realize the same level of recognition speed and recognition accuracy as those that are fixed in advance not only in how to write a character string but also in the line direction of the character string.

A frameless online character recognition device according to a second aspect of the present invention provides the character string information extracting means in the frameless online character recognition device according to the first aspect with the following determination function, The character string is written vertically based on the positional relationship between the baseline stroke detected by the baseline detection means and the character string in the handwriting according to the restriction on the positional relationship between the character string and the baseline. In addition to adding a discriminating function of judging whether the character is written horizontally or not, the rotation of the character string by the line correction means is not limited to the detection result of the line direction of the character string by the character string information extraction means, but also the vertical direction of the character string. It is characterized in that the determination is made by using the writing / horizontal writing discrimination result.

In such a configuration, although the positional relationship between the character string and the baseline is restricted, the character string is written while allowing the user to write and input the character string in an arbitrary manner and in an arbitrary direction. By simply writing a stroke indicating the inclination of the character string when writing, the line direction of the character string in the handwriting can be easily and automatically detected based on the stroke (base line). Whether the character string is written vertically or horizontally can be automatically detected based on the positional relationship. Thereby, the recognition speed and the recognition accuracy can be further improved.

The restrictions on the positional relationship between a character string and a baseline include, for example, writing a horizontal character string, drawing a baseline below the character string, or writing a vertical character string. Means that the baseline is drawn on the right side of the character string, and that the character string is written to the right if it is written horizontally, it is written down if it is written vertically, or the baseline is aligned with the line direction of the character string It is possible.

In addition, a stroke (baseline stroke) indicating the inclination of the character string itself is restricted, and a linear stroke written first, a linear stroke written last, a longest linear stroke, Alternatively, the stroke may be fixed in advance to any one of two linear strokes having a predetermined length or more, or at least two strokes may be set as candidates, and a stroke selected and designated by the user from the candidates. Accordingly, a baseline stroke can be easily detected from a plurality of strokes constituting a handwriting.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a frameless online character recognition device according to an embodiment of the present invention.

The frameless online character recognition device shown in FIG.
It comprises a handwriting input unit 11, a storage unit 12, a baseline detection unit 13, a character string information extraction unit 14, a line correction unit 15, and a character string recognition unit 16.

The handwriting input unit 11 is realized by a coordinate input device such as a tablet. The handwriting input unit 11 converts a handwriting written on a coordinate input surface (tablet surface) with a writing pen (not shown) into two-dimensional coordinates. Detect and input as time series.

The storage unit 12 is used to store a time series (handwriting coordinate sequence) of handwriting coordinates input by the handwriting input unit 11 and the like. The baseline detecting unit 13 detects a baseline stroke from a plurality of strokes indicated by the time series of handwriting coordinates stored in the storage unit 12.
Strokes other than the baseline stroke are regarded as character strings.

Here, the stroke means a handwriting from the time when the pen touches the coordinate input surface of the handwriting input unit 11 until the pen leaves. In addition, the baseline is a user (scribe)
Non-string part of the handwriting written by
A stroke indicating the inclination of a character string.

The character string information extraction unit 14 determines whether the line direction (character line direction) of the character string is to be written vertically or horizontally based on the positional relationship between the baseline detected by the baseline detection unit 13 and the character string. (Vertical writing / horizontal writing) is determined, and character string information indicating the result is output.

The line correcting unit 15 converts the handwriting coordinate sequence stored in the storage unit 12 based on the line direction indicated by the character string information output by the character string information extracting unit 14 and the distinction between vertical writing and horizontal writing. The line direction of the character string (coordinate sequence indicating) is corrected so that the upper, lower, left, and right of each character in the character string shown (in the handwriting) are the same as the upper, lower, left, and right coordinates.

The character string recognizing unit 16 performs a recognition process on a handwriting coordinate sequence of a character string obtained as a result of the line correction by the line correcting unit 15, and outputs a recognition result. Next, the operation of the configuration of FIG. 1 will be described with reference to FIGS.

First, it is assumed that the user has started writing with a pen on the coordinate input surface of the handwriting input unit 11. Then, the coordinate sequence (data) of the handwriting written by the user is detected by the handwriting input unit 11, and the coordinate sequence is stored in the storage unit 1.
2 are stored in chronological order. The data of each coordinate (coordinate data) is provided with an on / off flag indicating whether the pen is touching the coordinate input surface or separated from the coordinate input surface (whether the pen is separated from the coordinate input surface by a certain height or more within a range where the coordinates can be detected). ing.

In the present embodiment, the user specifies the line direction of a character string to be written on the coordinate input surface of the handwriting input unit 11 by drawing a linear stroke called a baseline. The following four types of strokes can be designated (used) as a baseline.

(A1) Of the linear strokes in the handwriting,
Stroke written first (a2) Of the linear strokes in the handwriting, the last stroke written (a3) Of the linear strokes in the handwriting, the longest stroke (a4) Of the linear strokes, a stroke having a length equal to or more than a certain length written twice is used by the user through a user interface (not shown) to determine which of the above four types of strokes to use as a baseline. Select and specify. When the user interface receives the specification of the selection of the baseline stroke type from the user, the user interface sets (the information on) the selected and specified baseline stroke type in the baseline detection unit 13. In the initial state, the baseline detection unit 13
In, a default baseline stroke type predetermined in the system is set. Therefore, when the selection of the baseline stroke type is not specified by the user, the stroke indicated by the default baseline stroke type is the stroke that can be specified as the baseline.

When the user completes the handwriting input of the character string, the baseline detecting unit 13 determines the baseline stroke (baseline stroke and baseline stroke) from the time series of handwriting coordinates (handwriting coordinate sequence) stored in the storage unit 12 at that time. The operation for detecting the stroke is performed as follows in accordance with the flowchart of FIG. The end of the character string input by the user can be determined by detecting a state in which the pen is away from the coordinate input surface for a certain period or more, that is, a state in which the coordinate data with the OFF flag is continued for a certain number or more. . In addition, an icon indicating the end of the handwriting input of the character string may be displayed, and the determination may be made by detecting that the user touches the icon with a pen.

First, the baseline detection unit 13 checks which of the above (a1) to (a4) the type of the baseline stroke set in the detection unit 13 (FIG. 14).
Step S1).

If the type (a1) of the baseline stroke (hereinafter referred to as type (a1)) is set, the baseline detection unit 13 determines the coordinates stored in the storage unit 12. Of the linear strokes in the handwriting indicated by the column, the stroke (coordinate sequence of) the first handwritten is detected as the baseline (step S2 in FIG. 14).

Further, the base line detecting section 13 performs the above (a2)
If the baseline stroke type (hereinafter referred to as type (a2)) is set, the stroke (coordinate sequence) of the last written stroke among the linear strokes in the handwriting is set as the baseline. (Step S3 in FIG. 14).

Further, the base line detecting section 13 performs the above (a3)
If the baseline stroke type (hereinafter referred to as type (a3)) is set, the stroke (coordinate sequence) of the longest stroke among the linear strokes in the handwriting is used as the baseline. (Step S4 in FIG. 14).

Here, the length of the stroke is, for example, the sum s of the lengths of the respective polygonal lines when the coordinate strings constituting the stroke are connected by polygonal lines. In addition, a linear stroke means that a ratio L / s of a length s of the stroke and a distance L between the start point and the end point of the stroke exceeds a predetermined threshold α as shown in the following equation. Refers to stroke.

Α <L / s ≦ 1 FIG. 2 illustrates the length of the stroke and the linear stroke. In the example of FIG. 2, the number of polygonal lines when connecting the coordinate sequence forming the stroke 21 with polygonal lines is 9, and the length of each polygonal line is s1, s2,..., S9.
It has become. In this case, the length s of the stroke 21 is
It is expressed by the following equation: s = s1 + s2 +... + S9.

Assuming that the distance between the start point 211 and the end point 212 of the stroke 21 is L, the value of L / s indicates the likeness of the stroke 21 as a straight line.
Is closer to a straight line.

Further, the base line detecting section 13 performs the above (a4)
If the type of the baseline stroke (hereinafter referred to as type (a4)) is set, of the linear strokes in the handwriting, the strokes of a certain length or more that are written twice or more (Coordinate sequence of) is detected as a baseline (step S5 in FIG. 14).

Here, the linear stroke written twice is one of the linear strokes in the handwriting shown in FIG.
The start points of two linear strokes 31 and 32 having a certain length or more are b1 and b2, and the end points are e1 and e2.
When the distance between any two points p and q is represented by D (p, q), the following conditions D (b1, b2) <γ and D (e1, e2) <γ or D (b1, e2) A stroke satisfying <γ and D (e1, b2) <γ. Here, γ is a predetermined threshold.

That is, a linear stroke written in double means that the distance between both start points and both end points of two linear strokes, or the start point and end point of each other is within the range of the distance indicated by γ. ) Means approaching each other.

When the baseline detecting section 13 detects a baseline stroke in any of the above steps S2 to S5, it notifies the character string information extracting section 14 of information on the stroke (step S6 in FIG. 14).

The character string information extraction unit 14 calculates the position of the base line (base line stroke) notified by the baseline detection unit 13 and the character string of the handwriting indicated by the coordinate sequence stored in the storage unit 12. , The line direction of the character string and the information of vertical writing / horizontal writing are detected.

Hereinafter, details of the detection of the line direction and the vertical / horizontal writing information of the character string by the character string information extracting unit 14 will be described. Note that the upper, lower, left, and right sides of the character string appearing in the following description are the upper and lower sides of the character string when viewed from the direction in which the upper, lower, left, and right of each character in the character string is correct. , Left and right.

FIG. 4 is for explaining the direction in which the character string is written. The direction of the horizontally written character string 41 is rightward, the direction of the horizontally written character string 42 is leftward, and the direction of the vertically written character string 43 is downward. The direction of the vertically written character string 44 is upward. In order for the character string information extraction unit 14 to determine whether the character string as shown in FIG. 4 is vertical writing or horizontal writing, it is necessary to detect the upper, lower, left and right of each character in the character string. There is. However, complicated processing is required to detect the top, bottom, left, and right of a character.

Therefore, in this embodiment, for the user, (b1) when writing a character string in horizontal writing, a baseline is drawn below the character string. (B2) When writing a character string in vertical writing, , By setting a constraint that a baseline is drawn on the right side of the character string, the position of the character string of the handwriting indicated by the coordinate sequence stored in the storage unit 12 and the position of the baseline (notified from the baseline detection unit 13) From the relationship, the character string information extraction unit 14 can easily determine whether the character string is written vertically or horizontally.

In addition, regarding the vertical and horizontal directions of the character string and the baseline that appear in the following description, the upper, lower, left and right directions when each character in the character string is viewed correctly in the upper, lower, left and right directions. Say. Therefore, in a character string written with the same inclination, there are two types of directions in each of vertical writing and horizontal writing.

FIG. 5 shows such an example. In the case of a horizontally written character string such as character strings 51 and 52, there are two types of directions, rightward and leftward. 5
As shown in FIG. 4, in the case of a vertically written character string, there are two types of directions, downward and upward. In addition, character strings 51 to 54
Is also drawn depending on whether the character string is written vertically or horizontally.
There are two types of directions (upward and downward for a vertically written character string and left and right for a horizontally written character string). In FIG. 5, the baseline BL is represented by a solid arrow, but this is to indicate the direction in which the baseline is written (drawn), and the baseline actually written by the user is simply a straight line. is there. This is the same for the baseline in other figures.

The character string information extracting unit 14 according to the present embodiment compares the character string (in the handwriting) indicated by the coordinate string stored in the storage unit 12 with the baseline (notified from the baseline detecting unit 13). To determine vertical / horizontal writing of the character string from the positional relationship, one of the following two types of vertical / horizontal writing determination methods (c1) and (c2) is applied.

First, when the vertical / horizontal writing discrimination method (c1) is applied, the character string information extraction unit 14 draws the character string of the handwriting on the character string as shown in the flowchart of FIG. If the baseline is horizontal and the character string appears to be written rightward, processing is performed to determine whether the character string is located on the upper or lower side of the baseline ( Step S11). Then, if the character string is located above the baseline, the character string information extraction unit 14 determines that the direction of the character string is horizontal writing (step S12). On the other hand, if the character string is located below the baseline, it is determined that the direction of the character string is vertical writing (step S13).

FIG. 9 shows a state in which the handwriting is viewed such that the base line is horizontal and the character string is rightward for such vertical / horizontal character string determination. In the example of FIG. 9, in the character strings 91 and 92 of the handwriting as shown in FIGS. 9A and 9B, the base lines BL1 and BL2 drawn with respect to the character strings 91 and 92 become horizontal.
In addition, when the character strings 91 and 92 appear to be written rightward, they are determined to be horizontal writing because they are located above the base lines BL1 and BL2.

Similarly, the character strings 93 and 94 of the handwriting as shown in FIGS. 9C and 9D have the base lines BL3 and BL4 drawn by the character strings 93 and 94, and the character strings 93 and 94 are horizontal. When it is seen that 93 and 94 are written rightward, they are determined to be vertical writing because they are located below the base lines BL3 and BL4.

On the other hand, when the vertical / horizontal writing discrimination method (c2) is applied, the character string information extracting unit 14 converts the character string of the handwriting into the character string as shown in the flowchart of FIG. When the drawn baseline is viewed as being horizontal and written rightward, a process is performed to determine whether the character string is located on the upper or lower side of the baseline (step S21). ). If the character string is located above the baseline, the character string information extraction unit 14 determines that the direction of the character string is horizontal writing (step S22). On the other hand, if the character string is located below the baseline, the direction of the character string is determined to be vertical writing (step S23).

FIG. 10 shows a state in which the handwriting is viewed such that the baseline is horizontal and rightward for the purpose of vertical / horizontal writing of such a character string. FIG.
In the example of 0, the character strings 101 and 102 of the handwriting as shown in FIGS. 7A and 7B are such that the base lines BL1 and BL2 drawn on the character strings 101 and 102 are horizontal and rightward. When viewed as written, it is determined to be horizontal writing because it is located above the baselines BL1 and BL2.

Similarly, the character strings 103 and 104 of the handwriting as shown in FIGS.
When the base lines BL3 and BL4 drawn on the line 04 are viewed horizontally and written rightward, they are located below the base lines BL3 and BL4.
It is determined as vertical writing.

Here, viewing the base line so as to be horizontal means that the starting point and the end point of the baseline are viewed horizontally, and there are two views.
FIG. 8 shows an example of these two viewpoints.
When the base line BL drawn in FIG. 1 is set to be horizontal, the base line BL is positioned below the character string 81 as shown in FIG.
There is a view that is located at the top of the character string 81 as shown in FIG.

When the character string in the vertical / horizontal discrimination method (c1) is viewed to be written rightward, as shown in the example of FIG. 9, the character string is viewed so that the baseline is horizontal. In this state, the start point of the character string is on the left and the end point is on the right.

When the baseline is written rightward in the vertical / horizontal discrimination method (c2), as shown in the example of FIG. 10, the baseline is horizontal. When viewed, the starting point of the baseline is on the left side, and the ending point is on the right side.

When the vertical / horizontal writing method of (c1) is used to determine the vertical / horizontal writing of the character string of the handwriting, that is, the handwriting is set so that the baseline is horizontal and the character string is rightward. In the case where vertical writing / horizontal writing is determined based on the positional relationship between the baseline and the character string in the viewed state, if there is no restriction on the direction in which the character string is written, FIG.
Of the character strings 61 to 68 shown in (a) to (h), the character strings 63 and 64 of the horizontal writing left direction and the character strings 67 and 68 of the vertical writing upward marked with X are erroneously determined.

Similarly, when the vertical / horizontal writing method of the above (c2) is used to determine the vertical / horizontal writing of the character string of the handwriting, that is, in a state where the handwriting is viewed so that the baseline is horizontal and rightward. If vertical / horizontal writing is determined based on the positional relationship between the baseline and the character string in the vertical direction, if there is no restriction on the direction in which the baseline is drawn, FIGS.
Of the character strings 71 to 78 shown in (h), a horizontally written character string 72 with a leftward baseline BL2, BL4 drawn.
74 and the vertically written character strings 76 and 78 on which the upward baselines BL6 and BL8 are drawn (character strings marked X in FIG. 7) are erroneously determined.

Therefore, in this embodiment, when the (d1) vertical / horizontal discrimination method (c1) is applied to the user,
If the character string is written horizontally, write it to the right, and if it is vertical, write it down. (D2) When applying the vertical / horizontal discrimination method (c2),
There is a restriction that the direction of the baseline is limited to one direction (here, the base line is directed rightward if the character string is written horizontally, and downward if the character string is written vertically).

However, the above-described restrictions are very natural when writing a character string vertically or horizontally, and do not impose a burden on the user. Here, a specific procedure of the vertical / horizontal character string discrimination processing (step S11 in FIG. 15) in the character string information extraction unit 14 according to the above-described discrimination method (c1) will be described with reference to the flowchart in FIG.

First, the character string information extraction unit 14
, A direction vector P from the start point to the end point of the character string portion, and a start point (xs, ys) of the base line from the character string of the handwriting stored in the character string and the coordinate line of the baseline (detected by the baseline detection unit 13). ) To the end point (xe, ye)
) Is obtained (step S3).
1) If the inner product (P, Q) of the direction vectors P and Q is (P, Q) <0, (xs, ys) and (xe, ye) are exchanged (steps S32 and S33).

Next, the character string information extraction unit 14 calculates the equation f (x, y) = (xe-xs) (y-ys) + (ys-) from the start point to the end point of the baseline.
ye) (x-xs) = 0 is obtained (step S34).

Next, the character string information extracting unit 14 obtains the coordinates (xg, yg) of the center of gravity of the coordinate string (xi, yi) (i = 1, 2,..., N) of the character string part according to the following equation ( Step S35).

Xg = (1 / N) Σxi, yg = (1 / N) Σyi where, Σxi and Σyi are i = 1 of xi and yi, respectively.
To i = N.

Next, the character string information extraction unit 14 determines in step S
Substituting the coordinates (xg, yg) of the center of gravity of the character string portion obtained in 35 into the above-mentioned equation f (x, y) of the straight line of the base line, and determining whether the base line and the character string are positive or negative by f (xg, yg) By examining the positional relationship with the center of gravity of the part, that is, the vertical and horizontal positional relationship between the baseline and the character string while looking at the handwriting so that the baseline is horizontal and the character string is rightward,
It is determined whether the corresponding character string is written vertically or horizontally (step S36).

Here, when f (xg, yg)> 0, it is determined that the character string is horizontal writing, and f (xg, yg)
If <0), it is determined that the character string is written vertically. Next, the above-described determination method in the character string information extraction unit 14
The specific procedure of the vertical / horizontal character string discrimination processing (step S21 in FIG. 16) according to (c2) will be described with reference to the flowchart in FIG.

The procedure here is the same as that in the vertical / horizontal writing discrimination process according to the discrimination method (c1) except that steps (S31 to S33) in the flowchart of FIG. 17 are unnecessary. The procedure is the same, and the vertical / horizontal writing of the character string is determined by the processing of steps S41 to S43 corresponding to steps S34 to S36 in the flowchart of FIG.

Next, a description will be given of a method of detecting the line direction of a character string in the character string information extraction unit 14. The method of detecting the line direction of this character string differs depending on whether the above (C1) or (c2) is applied as a vertical / horizontal writing discrimination method, and it is called (e1) when (C1) is applied. , When applying (C2), (e2)
Call. Note that, in the case of horizontal writing, the line direction of the character string means a direction in which the character string goes from left to right regardless of the written direction of the character string. Also, in the case of vertical writing, it means a direction in which a character string goes from top to bottom, regardless of the written direction of the character string.

FIG. 11 is for explaining the line direction of the character strings. The directions of the dotted arrows RD1 to RD4 attached to the character strings 111 to 114 correspond to the character strings 111 to 11 respectively.
4 shows the row direction.

When applying the vertical / horizontal discrimination method (c1), the character string information extraction unit 14 detects the line direction of the character string (e1).
According to 1), the direction of the vector from the point closer to the start point of the character string to the point closer to the end point of the character string between the start point and the end point of the baseline detected by the baseline detection unit 13 is represented by the character Row direction of column (step S in FIG. 15)
14). FIG. 12 shows an example of the line direction of a character string detected by this detection method (e1).
The direction of the dotted arrows RD1 to RD4 attached to 24 indicates the detected (character string) line direction. The solid arrows BL1 to BL4 indicate the baseline and its direction.

On the other hand, when the vertical writing / horizontal writing discrimination method (c2) is applied, the character string information extraction unit 14 determines the starting point of the baseline detected by the baseline detection unit 13 according to the character string line direction detection method (e2). The direction of the vector from to the end point is the line direction of the character string (step S2 in FIG. 16).
4). FIG. 13 shows an example of the line direction of a character string detected by this detection method (e2).
The direction of the dotted arrows RD1 to RD4 attached to 4 indicates the detected (character string) line direction. The solid arrow B
L1 to BL4 indicate a baseline and its direction.

The character string information extracting unit 14 determines the character string based on the positional relationship between the character string of the handwriting stored in the storage unit 12 and the baseline (detected by the baseline detecting unit 13) as described above. When the information on the row direction of the column and whether the character string is written vertically or horizontally is obtained, the information (character string information) is notified to the row correction unit 15 (step S15 in FIG. 15 or step S15 in FIG. 16). S2
5).

When the line correction unit 15 receives the character string information from the character string information extraction unit 14, it is stored in the storage unit 12 based on the line direction indicated by the character string information and the vertical / horizontal writing. The upper, lower, left, and right sides of each character of the character string of the handwriting are correctly positioned as in the example of FIG. 4 (specifically, vertical writing is parallel to the y-axis, and horizontal writing is parallel to the x-axis). As described above, a rotation conversion is performed on the coordinate sequence of the character string portion. The line correction unit 15 performs the rotation conversion on the coordinate sequence of the character string portion of the handwriting stored in the storage unit 12, that is, when the line direction of the character string is corrected, the character string recognition unit Notify 16.

Then, the character string recognizing section 16 changes the line correcting section 15
The character string portion subjected to the line correction according to the above is subjected to recognition processing while cutting out characters one by one according to whether the character string is written vertically or horizontally. Here, as a method of extracting characters from a character string, for example, “Online Character Recognition of Borderless Written Character String by Candidate Character Lattice Method” (Transactions of the Institute of Electronics and Communication Engineers Vol.J68-D No.4, p765, 1985) The described well-known cutting-out method may be used. Also,
As a method of character recognition, for example,
A well-known online character recognition method described in the issue may be used.

The function of each section of the above-described frameless online character recognition apparatus having the configuration of FIG. 1 is mainly performed by an information processing device represented by a pen input personal computer having a coordinate input surface such as a tablet or a portable information terminal. Baseline detector 13, character string information extractor 14, line corrector 15
And a storage medium such as a magnetic disk (floppy disk), an optical disk (CD-ROM, DVD, etc.), a semiconductor memory, etc., on which a program for functioning as the character string recognizing unit 16 is recorded. It is realized by mounting and reading and executing the program recorded in the storage medium by the information processing device.

In this way, the pen-input personal computer or portable information terminal recognizes handwritten handwriting input by hand on a coordinate input surface of a tablet or the like as a character, and operates the computer or the like based on the obtained information. It is possible to construct an online handwritten character input device which can efficiently input information into a computer and edit information input on the computer, has excellent portability, and is easy to handle even for a person who is unfamiliar with a keyboard.

The present invention is not limited to the above-described embodiment, and can be implemented in various modifications without departing from the scope of the invention. For example, in the above-described embodiment, the types of the baseline strokes that can be selected and designated by the user are four types (a1) to (a4), but are not limited thereto, and the available baseline strokes are fixed. It may be something that is.

In the above embodiment, the determination whether the character string is vertical writing or horizontal writing is made by applying the vertical writing / horizontal writing determination method (C1). Is performed based on the positional relationship between the baseline and the character string when the handwriting is viewed as if it were written rightward. In the vertical / horizontal discrimination method (C2), the baseline is horizontal, and It is based on the positional relationship between the baseline and the character string when looking at the handwriting as if it were written rightward, but this is not a limitation.For example, in the discrimination method (C1), The handwriting may be viewed as if the column is written leftward, and the handwriting may be viewed as if the baseline was written leftward in the determination method (C2). In this case, the positional relationship (upper / lower relationship) between the baseline and the character string is reversed between the vertical writing and the horizontal writing in the above embodiment.

In addition, the determination of the line direction of the character string is made when the handwriting is viewed as if the direction of the base line was vertical and the character string was written in a predetermined direction of upward or downward. The handwriting may be performed based on the positional relationship (left-right relationship) between the baseline and the character string, or as if the direction of the baseline is vertical and written in a predetermined direction, upward or downward. It may be performed based on the positional relationship (left-right relationship) between the baseline and the character string when viewed.

[0076]

As described in detail above, according to the present invention, while allowing a user to write and input a character string in an arbitrary writing style and in an arbitrary direction, a character string necessary for recognizing the character string is input. The detection of the line direction and the determination of the vertical / horizontal writing of a character string can be easily performed only by the user writing a stroke indicating the inclination of the character string when writing the character string, so that the recognition speed and the recognition accuracy are improved. .

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of a frameless online character recognition device according to an embodiment of the present invention.

FIG. 2 is an exemplary view for explaining a method of detecting a linear stroke in the embodiment.

FIG. 3 is an exemplary view for explaining a method of detecting a linearly written stroke in the embodiment;

FIG. 4 is a view for explaining a direction in which a character string is written, and shows a state in which each character string in the vertical and horizontal writing is viewed so that each character in the character string is vertically and horizontally correctly.

FIG. 5 is a view for explaining two types of directions of a baseline with respect to each character string of vertical writing and horizontal writing.

FIG. 6 is a diagram illustrating a state in which a handwriting is viewed such that a baseline is horizontal and a character string is directed rightward, for each of handwriting in which vertical writing / horizontal writing is correctly determined and handwriting which is erroneously determined;

FIG. 7 is a diagram illustrating a state in which a handwriting is viewed so that the baseline is horizontal and rightward, for each of handwriting in which vertical writing / horizontal writing is correctly determined and handwriting in which erroneous determination is made;

FIG. 8 is a view for explaining that there are two ways of looking at handwriting so that the baseline is horizontal.

FIG. 9 is a view for explaining the positional relationship between the baseline and the character string when the handwriting is viewed so that the baseline is horizontal and the character string is rightward.

FIG. 10 is a view for explaining a positional relationship between a baseline and a character string when the handwriting is viewed so that the baseline is horizontal and rightward.

FIG. 11 is a diagram for explaining the line direction of a character string.

FIG. 12 is a diagram showing an example of a character string detected by a detection method (e1) in the row direction.

FIG. 13 is a diagram showing an example of a character string detected by a detection method (e2) in the row direction.

FIG. 14 is a flowchart for explaining a baseline detection process by a baseline detection unit 13 in FIG. 1;

FIG. 15 is a flowchart for explaining character string information extraction processing including vertical / horizontal writing determination processing using a determination method (C1) by the character string information extraction unit 14 in FIG. 1;

FIG. 16 is a flowchart for explaining character string information extraction processing including vertical / horizontal writing determination processing using a determination method (C2) by the character string information extraction unit 14 in FIG. 1;

FIG. 17 is a flowchart for explaining a specific procedure of vertical / horizontal writing discrimination processing using the discrimination method (C1).

FIG. 18 is a flowchart for explaining a specific procedure of vertical / horizontal writing discrimination processing using the discrimination method (C2).

FIG. 19 shows the horizontal and vertical directions as viewed from the user, and the x axis and y.
The figure which shows the relationship with the direction of an axis.

[Explanation of symbols]

 11: Handwriting input unit, 12: Storage unit, 13: Baseline detection unit, 14: Character string information extraction unit, 15: Line correction unit, 16: Character string recognition unit, BL, BL1 to BL8: Baseline.

Claims (9)

[Claims] 1. A handwriting input unit for detecting and inputting a handwriting written on a coordinate input surface as a two-dimensional coordinate time series, and a plurality of strokes constituting the handwriting input by the handwriting input unit, the handwriting being input into the handwriting. Baseline detection means for detecting a baseline stroke that is a stroke indicating the inclination of the character string, character string information extraction means for detecting the line direction of the character string from the baseline stroke detected by the baseline detection means, A line correcting means for rotating the character string in the handwriting based on the detection result of the character string in the line direction by the character string information extracting means; and a character string for recognizing the character string rotated by the line correcting means. A frameless online character recognition device comprising: a recognition unit. 2. A handwriting input unit for detecting and inputting a handwriting written on a coordinate input surface as a time series of two-dimensional coordinates, and a plurality of strokes constituting the handwriting input by the handwriting input unit. A baseline detection means for detecting a baseline stroke which is a stroke indicating the inclination of the character string of the character string; and detecting a line direction of the character string from the baseline stroke detected by the baseline detection means, and determining a predetermined character. Based on the positional relationship between the line and the baseline, it is determined whether the character string is written vertically or horizontally based on the positional relationship between the detected baseline stroke and the character string in the handwriting. Character string information extracting means, a detection result of a character string in a row direction by the character string information extracting means, and a vertical / horizontal writing determination result of the character string Line-correcting means for rotating a character string in the handwriting based on the character string; and a character string recognizing means for recognizing the character string rotated by the line correcting means. Recognition device. 3. The method according to claim 1, wherein the baseline detection unit detects a first linear stroke as the baseline stroke from a plurality of strokes forming the handwriting. The online character recognition apparatus without a frame according to claim 2. 4. The method according to claim 1, wherein the baseline detection unit detects a last linear stroke as the baseline stroke from a plurality of strokes forming the handwriting. The online character recognition apparatus without a frame according to claim 2. 5. The apparatus according to claim 1, wherein the baseline detecting means detects a longest linear stroke as the baseline stroke from a plurality of strokes constituting the handwriting. Online character recognition device without the described frame. 6. A method according to claim 1, wherein said baseline detecting means detects, as the baseline stroke, a linear stroke having a predetermined length or more, which is written twice, from a plurality of strokes constituting the handwriting. 3. The frameless online character recognition device according to claim 1 or 2, wherein: 7. The system according to claim 1, wherein the first linear stroke, the last linear stroke, the longest linear stroke, and the double linear stroke having a predetermined length or more. 2. A stroke selected from at least two predetermined strokes of a linear stroke is detected as the baseline stroke from a plurality of strokes constituting the handwriting. Or the frameless online character recognition device according to claim 2. 8. A first step of detecting and inputting a handwriting written on a coordinate input surface as a time series of two-dimensional coordinates; and A second step of detecting a baseline stroke which is a stroke indicating the inclination of the middle character string, a third step of detecting a line direction of the character string from the baseline stroke detected in the second step, A fourth step of rotating the character string in the handwriting based on the detection result of the line direction of the character string in the third step; and a fourth step of recognizing the character string rotated in the fourth step. And a frameless online character recognition method. 9. A first step of detecting and inputting a handwriting written on a coordinate input surface as a time series of two-dimensional coordinates, and a plurality of strokes constituting the handwriting input in the first step, wherein the handwriting is obtained. A second step of detecting a baseline stroke, which is a stroke indicating the inclination of the character string in the middle, detecting a line direction of the character string from the baseline stroke detected in the second step, and According to the restriction on the positional relationship between the column and the baseline, it is determined whether the character string is written vertically or horizontally based on the positional relationship between the detected baseline stroke and the character string in the handwriting. A third step; rotation of the character string in the handwriting based on the detection result of the line direction of the character string and the result of vertical / horizontal writing determination of the character string in the third step; And a fourth step of recognizing the character string rotated in the fourth step.