AUTOMATIC LINE DIVISION SYSTEM FOR HANDWRITTEN TEXT FOR DIVISIONING LINES FROM AN ELECTRONIC HANDWRITING INPUT PANEL.

Abstract

An automatic line division system for handwritten text is proposed, designed to handle line division from an electronic handwriting input panel. During handwriting, various conditions typically arise, such as words that are too long or too short, isolated words, inverted strokes, etc. The system can address these issues based on a standard line height (10) and width. The edges and characteristics of the line division are detected, and specific values ​​are determined to correct the division and improve its efficiency. Furthermore, problems arising from splitting the first line are considered, and the issue of multiple lines due to incorrect line division can be corrected to a single line. These corrections enhance the accuracy of line division.Figure to be published with the abbreviation: Fig. 1.

Description

Title of the invention: AUTOMATIC LINE DIVISION SYSTEM FOR HANDWRITTEN TEXT FOR DIVISIONING LINES FROM HANDWRITTEN TEXT FROM AN ELECTRONIC HANDWRITING INPUT PANEL.

[0001] Domain of the invention.

[0002] The present invention relates to the division of handwritten text lines, and in particular an automatic handwritten text line division system for dividing handwritten text into lines from an electronic handwriting input panel.

[0003] Context of the invention.

[0004] Conventional writing on an electronic writing panel, handwritten texts are converted into computer texts, then arranged line by line.

[0005] Handwritten text division is a technology primarily used to separate words and numbers in handwritten texts. It is used in computer-based handwritten text recognition and to analyze these texts in order to provide support for machine learning and natural language processing. The basic principle involves analyzing the shapes, sizes, structures, continuities, and other characteristics of the handwritten texts. These characteristics are then used to recognize each word based on the identified features.

[0006] For handwritten text separation, current technologies widely employ deep learning processes, conventional computer visual sensitivity, and unsupervised trends. The deep learning-based process uses a convolutional neural network (CNN) with an end-to-end function to perform the separation work. In conventional computer visual sensitivity, comprehensive image processing technologies are used to extract numerous features from the handwritten texts, and these features are then used for classification and recognition based on figure separation. Unsupervised learning uses a set of images without any prior labeling or marking information, meaning that a single image is used to perform the text separation.

[0007] The main object of the present invention is to solve the problem of line separation for handwritten texts entered from electronic writing panels. In fact, for handwritten texts, the stroke order is not The order of the text in the handwritten text is equal to that of the authors' writing habits and the effects of the environment on writing. Furthermore, irregular strokes in handwritten texts all have an impact. However, these problems will lead to difficulties in separating the lines.

[0008] Summary of the invention.

[0009] Consequently, the object of the present invention is to provide an automatic line-splitting system for handwritten text, using an electronic handwriting input panel where, generally, various conditions arise during writing, such as words that are too long or too short, isolated words, inverted strokes, etc. The program of the present invention can address these problems in the writing process in order to improve the efficiency of line splitting. Furthermore, problems arising from splitting the first line are taken into consideration, for example, the problem of splitting several lines into one. These problems can be corrected by the methods of the present invention in order to improve the accuracy of line splitting.The present invention makes it possible to divide the lines of a handwritten text and to obtain better results than the methods used in prior arts.

[0010] To achieve the aforementioned objective, the present invention proposes an automatic line division system for dividing handwritten text from an electronic handwriting input panel; the system records the coordinates of each point of all strokes of the handwritten text and generates the time for each point; the direction of the lines being a linear direction (x-direction) and the different lines being arranged along a y-direction which is vertical with respect to the x-direction; the system comprising a central processing unit; the central processing unit comprising a processor and memory; the processor serving to perform the necessary operations; the memory serving to store the computer programs used and the data relating thereto; the system also comprising: a line height unit for determining a standard height and a standard width of the handwritten text;an anomaly removal unit connected to the line height unit, which receives the standard height and standard width from the line height unit; the anomaly removal unit is used to remove anomalous strokes in the handwritten text, an anomalous stroke being defined as a stroke height exceeding a defined time interval of the standard height, or a stroke width exceeding a defined time interval of the standard width; a pre-division unit connected to the anomaly removal unit to receive the handwritten text that has been processed by the anomaly removal unit; wherein the pre-division unit is used to roughly divide the handwritten text into multiple lines in advance along the y-axis direction; for; an original first line detected first, a leftmost edge of the original first line and a rightmost edge of this original first line are detected; then a height of a first set of times defined from the standard height is used as the line height in the division of the handwritten text, this line height is used to divide the handwritten text again in order to obtain a first line of the handwritten text; for the other parts of the handwritten text, the same process is repeated several times to divide the handwritten text into several lines; a special lines unit connected to the pre-division unit to receive the handwritten text after division by the pre-division unit; in this special lines unit, each isolation point and each inversion stroke is defined as a single line, which is incorporated into another adjacent line according to the following steps;an oversized line division unit connected to the special lines unit and the pre-division unit to receive the handwritten text processed by the special lines unit; the oversized line division unit is used to redivide the oversized lines that have been divided by the pre-division unit; The definition of an oversized line is as follows: for a line divided by the pre-division unit, if the height (in the y-axis direction) of the line is greater than a defined second time of the standard height, the line is considered an oversized line; a spaced embedding unit connected to the oversized line division unit to receive the handwritten text processed by the oversized line division unit; in which the spaced embedding unit serves to confirm that the division of the normal line is correct; if it is deemed incorrect, the normal line is discarded, while the original line before being processed by the oversized line division unit is accepted; lines formed by isolation points and inversion lines are also processed by the spaced embedding unit to be embedded in other lines;and a multi-line unit connected to the spaced embedding unit to receive the handwritten text from the spaced embedding unit; wherein the multi-line unit serves to determine the stroke distances of the lines of the spaced embedding unit; the distance is the distance between the rightmost end of a first stroke and the leftmost end of a second stroke; if the distance is greater than a specified width, the two strokes are considered boundaries, and the first and last strokes are divided into two separate lines; the handwritten text is divided into lines in the manner described above.

[0011] Brief description of the drawings.

[0012] Fig. 1 shows the connection of the elements of the present invention.

[0013] The [Fig.2] is a diagram of the operation of the present invention.

[0014] Fig. 3A illustrates an embodiment of the present invention and shows the distances between the points of a line and the y-axis.

[0015] Fig. 3B shows the distribution between the points of the strokes of a handwritten text with Dy.

[0016] Fig. 4 is a schematic view showing the directions of a handwritten text.

[0017] Figure 5 is a functional diagram showing the connection of the elements of the present invention. Detailed description of the invention

[0018] To enable those skilled in the art to better understand the present invention, a detailed description will be provided below. However, these descriptions and the accompanying drawings are intended only to assist those skilled in the art in understanding the objects, functions, and features of the present invention, but shall not be used to limit the scope and spirit of the present invention as defined in the attached claims.

[0019] With reference to Figures 1 to 5, in the present invention, handwritten texts refer to handwritten texts entered by a user from the electronic writing panels. The electronic writing panel 8 records each stroke (coordinates of the points forming the stroke) and the time of point generation. Consequently, the points forming each stroke and the time of point entry are recorded (step 110). In what follows, the line direction lies along the x-axis of a coordinate. The lines are arranged from one line to the next along the y-axis, as illustrated in [Fig. 4].

[0020] The structure of the present invention mainly comprises a central unit 100. The central unit 100 comprises a processor 2 and a memory 6. The processor 2 is used to perform the operations necessary for the present invention. The memory 6 is used to store the computer programs used in the present invention and the data related to the present invention. All results concerning the operation of the present invention and related data are stored in the memory 6.

[0021] A line height unit of 10 is used to arrange the strokes of a handwritten text in order according to stroke height, and a height selected from all heights based on a ratio setting of all heights is defined as a standard height. Similarly, the line height unit of 10 is used to arrange the strokes of a handwritten text in order according to stroke width, and a width selected from among the widths based on a ratio setting of all widths is defined as the standard width. For example, a handwritten text has 100 strokes, which are arranged in order according to stroke height, and the 89th height of a stroke in the order of heights is the standard height; the 89th width of a stroke in the order of widths is the standard width (step 115).

[0022] An anomaly removal unit 20 is connected to the line height unit 10, which receives the standard height and width from the line height unit 10. The anomaly removal unit 20 is used to remove abnormal strokes in handwritten texts (step 120), an abnormal stroke being defined as a stroke height greater than one standard height, or a stroke width greater than one standard width. For example, if a stroke height is more than 5 times the standard height, that stroke is removed.

[0023] A pre-division unit 30 is connected to the anomaly removal unit 20 to receive the handwritten texts that have been processed by the anomaly removal unit 20. The pre-division unit 30 serves to roughly divide the handwritten text into several lines in advance along the y-axis direction. For a first original line detected initially, the leftmost point of the first original line and the rightmost edge of this first original line are detected. Then, a height N1 times the standard height is used as the line height in the division of the handwritten text; this line height is used to divide the handwritten text again to obtain a first line of the handwritten text. For the other parts of the handwritten text, the same process is repeated several times to divide the handwritten text into several lines (step 125).

[0024] The pre-division condition will be described below.

[0025] Condition 1: a value obtained by subtracting the maximum value of the x-axis of all the strokes of a previous line from the maximum value of the x-axis of all the strokes of a current line is greater than the standard height;

[0026] Condition 2: a value obtained by subtracting the minimum value of the y-axis of a current treated stroke from the maximum value of the y-axis of all strokes in a current line is greater than twice the standard height; and

[0027] Condition 3: a minimum value of the y-axis of a current processed stroke is greater than a maximum value of the y-axis of a shift matrix, the shift matrix being defined as follows: for an X min of a minimum value of a current processed stroke in the x-axis and an X max of a maximum value of the current processed stroke in the x-axis, the shift matrix is ​​a matrix formed by all the strokes in the range (X min, X max), in the sense that a maximum value of these strokes in the y-axis is less than the minimum value of the current processed line in the y-axis.

[0028] If a feature satisfies condition 1 and condition 3, or satisfies condition 2 and condition 3, the feature currently being processed is considered to belong to a subsequent line.

[0029] For two continuous lines, determine the difference between the right end of an upper line of the two continuous lines and the left end of a subsequent line of the two continuous lines. This difference is used as the transverse difference dx. For two continuous lines, determine the difference between the lower end of the first word in an upper line of the continuous line and the upper end of the first word in a subsequent line of the continuous line. This difference is used as the longitudinal difference dy. The transverse difference dx must be greater than a defined time of the standard width, for example, 5 times. The longitudinal difference dy must be greater than zero. If the division of two lines meets these conditions, the division is correct. The larger the offset along the x-axis of the two lines, the more accurate the division.If dx and dy do not meet these conditions, the two continuous lines are considered to be in fact a single line.

[0030] A special line unit 40 is connected to the pre-division unit 30 to receive handwritten text after division by the pre-division unit 30. In this special line unit 40, each isolation point and each reversal stroke are defined as a single line (step 130), which is incorporated into another adjacent line in the following steps.

[0031] The isolation point is defined as follows: a distance along the y-axis Dy is obtained by subtracting a certain y-value of a stroke in a handwritten text from the y-value of a central point for all points of a stroke that are arranged along the y-axis. For all points of all strokes in handwritten texts, all Dy values ​​are determined, which are distributed according to a normal distribution, as illustrated in Figures 3A and 3B. For a stroke, if all points of that stroke lie on the outer edges of 95.449974% of all Dy values ​​(two standard deviations), that stroke is considered an isolation point and is defined as a single line.

[0032] In handwritten input, the time sequence of all entry points in the entry strokes is recorded and, consequently, the reversal stroke is defined as follows: for a defined stroke, the difference between the lowest point of the defined stroke on the y-axis and the highest point of a line containing the reversal stroke is greater than one N5 times the standard height, and the defined stroke is then called a reversal stroke.

[0033] An oversized line division unit 50 is connected to the special line unit 40 and the pre-division unit 30 to receive the handwritten text processed by the special line unit 40. The oversized line division unit 50 is used to redivide the oversized lines (step 135) which have been divided by the pre-division unit 30.

[0034] The definition of an oversized line is as follows: for a line divided by the pre-division unit 30, if the height (in the direction of the y-axis) is greater than N2 times the standard height, the line is considered to be an oversized line. A grouping algorithm (such as the k-means algorithm) is then used to perform the grouping operation based on the y-axis height of each stroke. All strokes are grouped according to their y-axis height. The results are then analyzed for each stroke or word. Thus, excessively wide strokes can be divided into several normal strokes. The subdivided normal strokes are then rearranged according to the order of the handwritten text (step 140).

[0035] A spaced embedding unit 60 is connected to the oversized line division unit 50 to receive the handwritten text processed by the oversized line division unit 50. The spaced embedding unit 60 is used to determine the transverse difference dx and the longitudinal difference dy for the upper normal lines after processing by the oversized line division unit 50, in order to determine whether the division of the normal line is correct. If it is considered incorrect, the normal line is discarded, while the original line previously processed by the oversized line division unit 50 is accepted (step 145).

[0036] Indeed, the lines formed by the isolation points and the inversion lines are also processed by the spaced incorporation unit 60 to be incorporated into other lines (step 140).

[0037] A multi-line unit 70 is connected to the spaced embedding unit 60 to receive the handwritten text from the spaced embedding unit 60. The multi-line unit 70 is used to determine the stroke distances of the lines in the spaced embedding unit 60. The distance is the distance between the rightmost end of a previous stroke and the leftmost end of a last stroke. If the distance is greater than a defined width N7, the two strokes are considered boundaries, and the first and last strokes are divided into two separate lines (step 150).

[0038] A whole-word fitting unit 80 is connected to the line-height unit 10 to receive the standard height and standard width of the line-height unit 10. The whole-word fitting unit 80 then determines whether the handwritten text consists of multiple lines (step 155). If it consists of multiple lines, the result of the above handwritten text division is accepted (step 160). If there are no multiple lines, it is determined to be a single word (step 165). The whole-word fitting unit 80 determines the height and width of the handwritten text. If the height of the handwritten text is less than N3 times the standard height, or if the width is less than N4 times the standard width, the whole handwritten text is considered to be a single word. A ratio between the width and height of the handwritten text can also be used as the fitting standard.For example, if the ratio between width and height is less than 2, the entire handwritten text is considered to be . a single word (step 170). Otherwise, the entire handwritten text is constructed from a single line (step 175).

[0039] A test unit is used to test other handwritten texts.

[0040] In practice, the system determines the standard height and width. Abnormal features are detected and removed to prevent them from affecting word division and recognition. The pre-division unit is used to roughly divide the handwritten text into several lines. Isolation points and inversion lines are defined in advance as single lines. Oversized lines are then divided into several normal lines. The normal lines must be arranged, and spaces must be incorporated. If the distances between words are too large, they will be further divided into several lines to improve division accuracy. It then checks whether the entire handwritten text constitutes a single word to determine if another division process is necessary.

[0041] According to features of the invention:

[0042] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel, further includes a whole word adjustment unit which is connected to the line height unit to receive the standard height and standard width of the line height unit, and then the whole word adjustment unit determines whether the handwritten text is a single line.

[0043] The whole word adjustment unit determines the height and width of the handwritten text; if the height of the handwritten text is less than one-fourth of the standard height or if the width is less than one-fifth of the standard width, the whole handwritten text is considered to be only one word; a ratio between the width and height of the handwritten text can also be used as a standard for adjustment.

[0044] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel, wherein the line height unit is used to arrange the strokes of handwritten text in order based on stroke heights, and a height selected from all heights based on a setting ratio of all heights is defined as a standard height, the line height unit is used to arrange the strokes of handwritten text in order based on stroke widths, and a width selected from widths based on a setting ratio of all widths is defined as a standard width.

[0045] The automatic line division system for handwritten text, for dividing lines of handwritten text from an electronic handwriting input panel, wherein the pre-division conditions are as follows:

[0046] - Condition 1: a value obtained by subtracting the maximum value of the x-axis of all the strokes of a previous line of the maximum value of the x-axis of all the strokes of a current line is greater than the standard height;

[0047] - Condition 2: a value obtained by subtracting the minimum value of the axis of the ordinates of a current treated stroke, the maximum value of the ordinate axis of all strokes in a current line is greater than twice the standard height; and

[0048] - Condition 3: a minimum value of the y-axis of a currently processed feature is greater than a maximum value of the y-axis of a shift matrix, the shift matrix being defined as follows: for an X min of a minimum value of a current processed stroke in the x-axis and an X max of a maximum value of the current processed stroke in the x-axis, the shift matrix is ​​a matrix formed by all strokes in the range (X min, X max), in the sense that a maximum value of these strokes in the y-axis is less than the minimum value of the current processed stroke in the y-axis.

[0049] - If a feature satisfies condition 1 and condition 3, or satisfies condition 2 and under condition 3, we consider that the feature currently being processed belongs to a following line.

[0050] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel in which, for two continuous lines, the determination of a difference between the rightmost end of an upper line of the two continuous lines and the leftmost end of a subsequent line of the two continuous lines, which is used as the transverse difference dx; for two continuous lines, the determination of a difference between the lower end of a first word in an upper line of the continuous line and the upper end of a first word in a subsequent line of the continuous line, which is used as the longitudinal difference dy; the transverse difference dx must be greater than a defined time of the standard width, for example 5 times, the longitudinal difference dy must be greater than zero;If the division of two lines meets these conditions, the division is correct; the larger the offset on the x-axis of two lines, the more precise the division; if dx and dy do not meet these conditions, the two continuous lines are considered to be in fact a single line.

[0051] The automatic line division system for handwritten text, for dividing lines of handwritten text from an electronic handwriting input panel, wherein the isolation point is defined as follows: a distance along the y-axis Dy is obtained by subtracting a certain y-value of a stroke in a handwritten text from the y-value of a central point for all points of a stroke that are arranged along the y-axis; for all points of all strokes in handwritten texts, all the D y values ​​are determined, which are distributed as a distribution normal; for a stroke, if all points of that stroke are on the outside sides of 95.449974% of all Dy (two standard deviations), that stroke is considered an isolation point and is defined as a single line.

[0052] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel, wherein the reversal stroke is defined as follows: for a defined stroke, the difference between the lowest point of the defined stroke on the y-axis and the highest point of a line containing the reversal stroke is greater than one-sixth of the standard height, and the defined stroke is then called a reversal stroke.

[0053] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel, wherein a grouping algorithm is used to perform the grouping operation on the basis of the y-axis height of each stroke; all strokes are grouped on the basis of their height on the y-axis; the results are then analyzed for each stroke or word; the oversized lines are thus subdivided into several normal lines; the subdivided normal lines are rearranged on the basis of the order of the handwritten texts.

[0054] The automatic line division system of a handwritten text for dividing lines of a handwritten text from an electronic handwriting input panel according to the invention, in which the normal lines are reorganized on the basis of the orders in the handwritten texts.

[0055] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein the spaced embedding unit serves to determine the transverse difference dx and the longitudinal difference dy for the upper normal lines after processing by the oversized line division unit in order to determine whether the division of the normal line is correct; if it is considered incorrect, the normal line is discarded, while the original line before processing by the oversized line division unit is accepted.

[0056] The advantages of the present invention lie in the fact that, generally, during writing, various conditions arise, such as words that are too long or too short, isolated words, reversed strokes, etc. The program of the present invention can address these problems in writing in order to improve the efficiency of line division. Furthermore, problems arising from the initial line division are taken into consideration, for example, the problem of dividing several lines into one. These problems can be corrected by the methods of the present invention in order to improve the accuracy of line division. The present invention makes it possible to dividing the lines of a handwritten text and achieving better results than the processes used in previous arts.

[0057] According to different possible embodiments of the invention:

[0058] The automatic line division system for handwritten text, for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, further comprises a whole-word adjustment unit which is connected to the line height unit to receive the standard height and standard width from the line height unit, and then the whole-word adjustment unit determines whether the handwritten text is a single line; and

[0059] The whole word adjustment unit determines the height and width of the handwritten text; if the height of the handwritten text is less than one-fourth of the standard height or if the width is less than one-fifth of the standard width, the whole handwritten text is considered to be only one word; a ratio between the width and height of the handwritten text can also be used as a standard for adjustment.

[0060] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein the line height unit is used to arrange the strokes of handwritten text in order based on the stroke heights, and a height selected from all the heights based on a setting ratio of all the heights is defined as a standard height, the line height unit is used to arrange the strokes of handwritten text in order based on the stroke widths, and a width selected from the widths based on a setting ratio of all the widths is defined as a standard width.

[0061] [The automatic line division system for handwritten text, for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein the pre-division conditions are as follows:

[0062] - Condition 1: a value obtained by subtracting the maximum value of the x-axis of all the strokes of a previous line of the maximum value of the x-axis of all the strokes of a current line is greater than the standard height;

[0063] - Condition 2: a value obtained by subtracting the minimum value of the axis of the ordinates of a current treated stroke, the maximum value of the ordinate axis of all strokes in a current line is greater than twice the standard height; and

[0064] - Condition 3: a minimum value of the y-axis of a currently processed feature is greater than a maximum value of the y-axis of a shift matrix, the shift matrix being defined as follows: for an Xmin of a minimum value of a current processed feature in the x-axis and an Xmax of a maximum value of the current processed feature in the x-axis, the matrix shift is a matrix formed by all the strokes in the range (X min, X max), in the sense that a maximum value of these strokes in the y-axis is less than the minimum value of the current processed line in the y-axis;

[0065] - if a feature satisfies condition 1 and condition 3, or satisfies condition 2 and under condition 3, we consider that the feature currently being processed belongs to a following line.

[0066] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein, for two continuous lines, the determination of a difference between the rightmost end of an upper line of the two continuous lines and the leftmost end of a subsequent line of the two continuous lines, which is used as the transverse difference dx; for two continuous lines, the determination of a difference between the lower end of a first word in an upper line of the continuous line and the upper end of a first word in a subsequent line of the continuous line, which is used as the longitudinal difference dy; the transverse difference dx must be greater than a defined time of the standard width, for example 5 times, the longitudinal difference dy must be greater than zero;If the division of two lines meets these conditions, the division is correct; the larger the offset on the x-axis of two lines, the more precise the division; if dx and dy do not meet these conditions, the two continuous lines are considered to be in fact a single line.

[0067] Automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein the isolation point is defined as follows: a distance on the y-axis Dy is obtained by subtracting a certain y-value of a stroke in a handwritten text from the y-value of a central point for all points of a stroke that are arranged along the y-axis; for all points of all strokes in handwritten texts, all the D y are determined, which are distributed as a normal distribution; for a stroke, if all points of that stroke are on the outside sides of 95.449974% of all Dy (two standard deviations), that stroke is considered an isolation point and is defined as a single line.

[0068] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein the reversal stroke is defined as follows: for a defined stroke, the difference between the lowest point of the defined stroke on the y-axis and the highest point of a line containing the reversal stroke is greater than one-sixth of the standard height, and the defined stroke is then called a reversal stroke.

[0069] The automatic line division system for handwritten text for dividing lines of handwritten text from an electronic handwriting input panel according to the invention, wherein a grouping algorithm is used to perform the grouping operation on the basis of the y-axis height of each stroke; all strokes are grouped on the basis of their height on the y-axis; the results are then analyzed for each stroke or word; the oversized lines are thus subdivided into several normal lines; the subdivided normal lines are reorganized on the basis of the order of the handwritten texts.

[0070] The automatic line division system of a handwritten text for dividing lines of a handwritten text from an electronic handwriting input panel according to the invention, in which the normal lines are reorganized based on the orders in the handwritten texts.

[0071] The automatic line division system of a handwritten text for dividing lines of a handwritten text from an electronic handwriting input panel according to the invention, wherein the spaced embedding unit serves to determine the transverse difference dx and the longitudinal difference dy for the upper normal lines after processing by the oversized line division unit in order to determine whether the division of the normal line is correct; if it is considered incorrect, the normal line is discarded, while the original line before processing by the oversized line division unit is accepted.

[0072] The present invention is thus described, but it is evident that it can be modified in many ways. Such variations should not be considered a departure from the spirit and scope of the invention, and all modifications that would be obvious to a person skilled in the art are intended to be included within the scope of the following claims.

Claims

1. Claims An automatic handwritten text line division system for dividing lines of handwritten text from an electronic handwriting input panel; the system records the coordinates of each point of all strokes of the handwritten text and generates the writing time for each point; the direction of the lines being a linear direction (x-direction) and the individual lines being arranged along a y-direction which is vertical to the x-direction; the system comprises a central processing unit; the central processing unit comprises a processor (2) and a memory (6); the processor (2) is used to perform the necessary operations; the memory is used to store the computer programs used and the corresponding data; the system further comprising: - a line height unit (10) for determining a standard height and a standard width of a line of the handwritten text; - an anomaly removal unit (20) connected to the line height unit, which receives the standard height and the standard width from the line height unit; the anomaly removal unit is used to remove abnormal strokes in the handwritten text, the abnormal stroke being defined by the height of a stroke being greater than one times the standard height, or the width of a stroke being greater than one times the standard width; - a pre-division unit (30) connected to the anomaly removal unit for receiving the handwritten text which has been processed by the anomaly removal unit; wherein the pre-division unit is for approximately dividing the handwritten text into multiple lines in advance along a y-axis direction; for a first original line detected first, a leftmost edge of the first original line and a rightmost edge of this first original line are detected; then, a first-time set height of the standard height is used as a line height in dividing the handwritten text, this line height is used for dividing the handwritten text again so as to to obtain a first line of the handwritten text; for the other parts of the handwritten text, we proceed in the same way to divide the handwritten text into several lines; - a special line unit connected to the pre-division unit for receiving the handwritten text after division by the pre-division unit; in this special line unit, each isolation point and each inversion stroke are defined as a single line, which is incorporated into another adjacent line in the following steps; - an oversize line division unit connected to the special line unit and the pre-division unit to receive the handwritten text processed by the special line unit; the oversize line division unit is used to redivide the oversize lines that have been divided by the pre-division unit; the definition of the oversize line is as follows: for a line divided by the pre-division unit, if the height (in the y-axis direction) of the line is greater than a second series of times the standard height, the line is considered an oversize line; - a spaced embedding unit connected to the oversized line division unit for receiving the handwritten text processed by the oversized line division unit, the spaced embedding unit being responsible for asserting that the division of the normal line is correct; if it is considered incorrect, the normal line is discarded, while the original line before being processed by the oversized line division unit is accepted; for the lines formed by the isolation points and the inversion lines are also processed by the spaced embedding unit to be incorporated into other lines; - a multiple line unit connected to the spaced embedding unit for receiving handwritten text from the spaced embedding unit; wherein the multiple line unit is used to determine the distances between strokes of the lines of the spaced embedding unit; the distance is the distance between the rightmost end of a first stroke and the leftmost end of a last stroke; if the distance is greater than a determined width, the two strokes are considered as boundaries, and the first and last strokes are divided into two different lines; and - the handwritten text is divided into lines in the manner described above.

2. An automatic handwritten text line division system for dividing lines of handwritten text from an electronic handwriting input panel according to claim 1, further comprising a whole-word adjustment unit which is connected to the line height unit to receive the standard height and the standard width of the line height unit, and then the whole-word adjustment unit determines whether the handwritten text is a single line; and the whole-word adjustment unit determines the height and width of the handwritten text; if the height of the handwritten text is only one-fourth times the standard height or the width is only one-fifth times the standard width, the whole-word text is considered to be only one word; a ratio of the width to the height of the handwritten text may also be used as a standard for adjustment.

3. An automatic handwritten text line division system for dividing lines of handwritten text from an electronic handwriting input panel according to claim 1, wherein the line height unit is for arranging strokes of handwritten text in order based on the heights of the strokes, and a height selected from all the heights based on an adjustment ratio of all the heights is defined as a standard height, the line height unit is for arranging strokes of handwritten text in order based on the widths of the strokes, and a width selected from the widths based on an adjustment ratio of all the widths is defined as a standard width.

4. An automatic handwritten text line division system for dividing handwritten text lines from an electronic handwriting input panel according to claim 1, wherein the pre-division conditions are as follows: - Condition 1: A value obtained by subtracting the maximum x-axis value of all strokes of a previous line from the maximum x-axis value of all strokes of a current line is greater than the standard height; - Condition 2: A value obtained by subtracting the minimum y-axis value of a current processed stroke from the maximum y-axis value of all strokes in a current line is greater than twice the standard height; and - Condition 3: A minimum y-axis value of a current processed stroke is greater than a maximum y-axis value of a shift matrix, the shift matrix being defined as follows: for an X min of a minimum value of a current processed stroke in the x-axis and an X max of a maximum value of the current processed stroke in the x-axis, the shift matrix is ​​a matrix formed by all strokes in the range (X min, X max), in the sense that a maximum value of these strokes in the y-axis is less than the minimum value of the current processed line in the y-axis;- if a line satisfies condition 1 and condition 3, or satisfies condition 2 and condition 3, the line currently being processed is considered to belong to a following line.;

5. An automatic handwritten text line division system for dividing lines of handwritten text from an electronic handwriting input panel according to claim 1, wherein, for two continuous lines, determining a difference between the rightmost end of an upper line of the two continuous lines and the leftmost end of a next line of the two continuous lines, which is used as a transverse difference dx; for two continuous lines, determining a difference between the lower end of a first word in an upper line of the continuous line and the upper end of a first word in a next line of the continuous line, which is used as a longitudinal difference dy; the transverse difference dx must be greater than a set time of the standard width, e.g., 5 times, the longitudinal difference dy must be greater than zero;if the division of two lines meets these conditions, the division of two lines is correct; the greater the shift on the x axis of two lines, the more accurate the division; if dx and dy do not meet these conditions, the two continuous lines are considered to be actually a single line.;

6. An automatic handwritten text line splitting system for splitting handwritten text lines from an electronic handwriting input panel according to claim 1, in which the isolation point is defined as follows: a distance on the y-axis Dy is obtained by subtracting a certain y-value of a stroke in a handwritten text from the y-value of a central point for all points of a stroke that are arranged along the y-axis; for all points of all strokes in handwritten texts, all D y are determined, which are distributed as a normal distribution; for a stroke, if all points of this stroke lie at the outer sides of 95.449974% of all Dy (two standard deviations), this stroke is considered an isolation point and is defined as a single line.

7. An automatic handwritten text line division system for dividing handwritten text lines from an electronic handwriting input panel according to claim 1, wherein the reversal stroke is defined as follows: for a defined stroke, the difference between the lowest point of the defined stroke on the y-axis and the highest point of a line containing the reversal stroke is greater than one-sixth of the standard height, and the defined stroke is then called a reversal stroke.

8. An automatic handwritten text line division system for dividing handwritten text lines from an electronic handwriting input panel according to claim 1, wherein a clustering algorithm is used to perform the clustering operation based on the y-axis height of each stroke; all strokes are clustered based on their y-axis height; the results are then analyzed for each stroke or word; the oversized lines are thus redivided into multiple normal lines; the redivided normal lines are rearranged based on the order of the handwritten texts.

9. An automatic handwritten text line division system for dividing handwritten text lines from an electronic handwriting input panel according to claim 1, wherein the redivided normal lines are rearranged based on the orders in the handwritten texts.

10. An automatic handwritten text line division system for handwritten text line division from an electronic handwriting input panel according to claim 5, wherein the spaced embedding unit is for determining the transverse difference dx and the longitudinal difference dy for the upper normal lines after processing by the oversize line division unit to determine whether the division of the normal line is correct; if it is considered incorrect, the normal line is discarded, while the original line before processing by the oversize line division unit is accepted.

Images