CZ6196A3 - Method of segmentation of draws for hand-writing input - Google Patents

Abstract

The method of the present invention includes a step of calculating the derivative (140), or instantaneous rate of change, of the curvature at points in the handwritten input (110). The method then selects as stroke boundary points certain points (or pixels) in the input which lie at a midpoint between a point of high curvature derivative and a succeeding point of low curvature derivative (150). Such boundary points are not influenced by absolute curvature values, but rather only by relative changes in the curvature. The stroke segmentation boundary points are provided to a stroke-based recognizer for interpretation of the handwritten input (170).

Description

Method of stroke segmentation for handwritten input

Technical field

LO σ>

rc

The present invention relates generally to the art of tapigraphy and, more particularly, to a method of stroke segmentation for handwriting

7SÍIÍÍ

BACKGROUND OF THE INVENTION

Machine affair and resources with the task of taking different approaches to splitting handwriting strokes'

UZltl recognition of human handwriting is very complex with the simultaneous rapid expansion of computational pen input options has become important to be addressed. There are a large number of its solutions. One useful approach is to enter a sequence of basic moves, or these moves, which are parameterized in some way, as inputs of a character recognizer.

A key condition for a stroke-based recognizer is that many instances of the same character class (eg, the letter A written at different times by different writers) should always be divided into a similar set of strokes. This helps to ensure that recognition is not too complicated, since character case descriptions will appear similar to character recognition. Ideally, all instances of a given feature would always contain the same number of strokes, all strokes would be in the same relative position, and the stroke property descriptions would be very similar in all cases. This ideal is not achievable in practice, but the accuracy of recognition can be improved to the extent that it can be approached.

_______ In one ______ of — yet-used — methods, the boundaries of the jtati are located at points where the speed of the pen in the vertical (or y) direction is zero, i. at points where the font begins to move up or down. The resulting stroke sets can therefore be called upstrokes ”and downstrokes. This method is discussed in Mermelstein & Eden, Experiments on Computer Recognition of Connected Handwritten Vords, in Information and Control, vol.7, pp.255-270, 1964. One of the drawbacks of this method is that it is too sensitive to changes in the vertical direction and insensitive to changes in the horizontal direction. However, many characters are composed of horizontal parts - for example, the horizontal line t and the three legs E are usually much more horizontal than vertical, even in a sloppy manuscript. The y-speed-based stroke segmentator sometimes divides the horizontal portion into one stroke, but often divides it into two, three, or more strokes, simply because of the small vibration of the pen in the vertical direction. This leads to low recognition accuracy, since many instances of the same feature will often be segmented into different looking stroke sets. Attempts to correct the inaccuracies of this method, including the requirement of minimal change in the vertical direction before creating a new thrust, have had limited success and many of the same underlying problems still remain.

Another current method solves this problem by placing stroke boundaries at points where the curvature is at its local ₍ maximum and exceeds a certain threshold value that corresponds to sharp bends in the manuscript. Since sharp bending may occur regardless of the direction in which the pen moves) this method is sensitive to the orientation of different parts of handwriting, such as words or characters, but the technique of curvature also has its drawbacks, for example, suppose someone writes L · with a very pronounced curvature rather than a sharp curvature, so it starts to look more In this case, the method may fail on segmentation L if the threshold curvature required for stroke boundaries has not been reached. Simply lowering the threshold does not solve the problem, as this would result in an excessive number of strokes. like their 'too small' count, or This again means that many cases of the same character are often segmented into different types of strokes.

Thus, a segmentation method is needed which is more accurate and does not have the drawbacks of said methods, such as the speed direction method and the existing curvature method.

SUMMARY OF THE INVENTION

In accordance with the present invention, a method of segmenting a handwriting input into a number of strokes is implemented

In consistent for many cases, each individual character class.

the present invention provides a method of segmenting handwritten entry into strokes that are similar in shape and location for many cases of each individual character class at the entry.

The method of the present invention generally comprises the step of calculating the derivative or instantaneous value of the curvature change at the handwriting entry points. The method then selects certain points (or pixels) in the input that lie between the high derivative of the curvature and the subsequent low derivative of the curvature as the stroke boundary points. Such boundary points are not affected by absolute curvature values, but rather only by relative changes in curvature.

_______ Overview of the drawings _

Fig. 1 shows a flow chart of a procedure for identifying stroke boundaries in accordance with a preferred embodiment of the present invention.

Giant. 2 shows segmentation handwriting entry to moves created before speed at y direction Fig.3 created displays segmentation in the previous way handwriting curvature. entry to moves Giant. 4 segmentation handwriting entry to moves

produced by a preferred embodiment of the present invention.

Figure 5 shows the points forming the letter L as received from the digitizer.

Fig. 6 shows points forming the letter L after resampling at a constant distance in accordance with a preferred embodiment of the present invention.

Fig. 7 is an exploded view showing the curvature calculation in a preferred embodiment of the present invention.

Figure 8 graphically shows the curvature values calculated for each point in Figure 7.

Figure 9 graphically depicts derivative values calculated for each point of Figure 7.

curvature

DETAILED DESCRIPTION OF THE INVENTION

Typically, handwritten character input is obtained from the user in the form of separate continuous segments. A separate continuous segment consists of one or more pen strokes, where the pen stroke is the trace that the pen leaves 'during' its contact time - with the input device - such as a digitizing tablet or paper.

In the present invention, the recognizable handwriting input units are formed by one or more clad continuous segments. Handwriting input is an input that is scanned electronically, including but not limited to: handwriting input; electronic input; a pressure sensed input, such as a print input; input received electronically, such as by fax, radio search system receiver, or other device.

The stroke is expressed as a sequence of points sampled at approximately regular intervals by the input device. Each point is determined at least by x and y coordinates. The strokes may be scanned electronically using a digitizing tablet or otherwise may be derived from a scanned or faxed image by a t-line image detection process; such electronic scanning methods are known in the art. In a preferred method, handwriting input is received by a device such as a personal digital assistant (PDA) or other device. Other means for receiving handwritten input include, but are not limited to: computers, modems, radio search system receivers, telephones, digital televisions, interactive televisions, digitizing tablet devices, fax machines, scanning devices, and other devices capable of capture handwriting input. Generally, when strokes are sensed electronically, each point is represented by an i-image element such that the stroke is expressed as a series of pixels on the device.

In accordance with the present invention may be handwritten input in the form of alphanumeric characters, ideographic characters or other forms of j characters or symbols of written communication ^-. '

2 and 3 of the handwriting input in the interpretation of the input, the recognizers based on the same alphanumeric input as in represent the alphanumeric stroke segmentation with a high probability of inaccuracy when stroke segmentation comes into strokes. Fig. 4 shows stroke segmentation Stroke segmentation performed in accordance with the present invention, and such stroke segmentation has a high probability of accurate interpretation as it arrives at the recognizer based on strokes.

Fig. 1 shows a flow chart of a preferred method in accordance with the present invention. Handwriting input from a digitizer or other device is received in the form of x and y coordinates 110 (with associated pen raised or lowered states). Typically, these dots are represented by pixels. The present method generally resample handwritten input to obtain points that are equally spaced over the length of handwritten input 120. FIG. 5 shows an example of an L 500 as a series of dots or pixels prior to resampling. Fig. 6 shows the same letter " L 600 as a series of dots or pixels after resampling. Resampling is done using a point-to-point distance d 610 that is constant throughout the handwriting input. Preferably, the value of d is selected such that the mean value of the handwriting input height is approximately between 15 and 30 times d. For example, in Figure 6, the value of d is selected so that the mean value of the letter height in the word is approximately 15 to 30 times d. .

The preferred embodiment in Fig. 1 calculates the curvature at each resampling point 130. Fig. 7 illustrates graphically the description of the data for calculating the curvature at point R 710. The curvature at the resampled point

Ί

R 710 is defined as the distance to the point following R (point R + 1730) from the point P 720. obtained by linear projection from the point of the upstream r-R (R-T; 750) through its own R; This distance is shown in Figure 7 as element 740. The curvature at the endpoints of handwritten input is defined as the same

in the corresponding-neighbor-neighbors. points -------- Curvature. at the internal points of the handwritten entry it can also be calculated by projecting two points distant from R, rather than one point • f, using a point that is two points centered R rather than point F.-1). to get a more robust error estimate. Fig. 8 shows graphically the curvature values of the obtained probe shown in Fig. 7:

For example, the aforementioned example of an L with a pronounced curvature can be segmented into vertical and horizontal strokes if the two straight L portions are substantially more straight than the bend between them. Thus, the curvature would increase towards the bend (i.e., the derivative of the curvature would be high) and decrease from the bend (the derivative of the curvature would be low), allowing the bending stroke to be at or near the bend as desired.

If, in a preferred embodiment of the present invention, the curvature for each resampling point has already been obtained, the curvature group for the resampling points may be smoothed to minimize any known artifacts introduced by the digitization process. The type of smoothing to be performed should be the standard method that is selected based on the particular digitizing features. This may include a point with its adjacent points (weighting of the point of higher points) and replacing the curvature value at said point with a calculated average. Ideally, the size of the anti-aliasing window used herein should be wider in areas of low curvature of the font and narrower in areas of high curvature to minimize the loss of important information when smoothing averages the value of the process itself. Since the object of smoothing the curves itself, one preferred smoothing method is to calculate the initial curves where smoothing is based on these curves, and then re-smoothing based on the newly calculated curves.

In a preferred embodiment of the present invention, an absolute curvature value is calculated for each resampling point by multiplying each negative curvature value by -1. Absolute curvature values are used for calculating curvature derivatives rather than actual curvature values, since the overall preferred embodiments of the present invention deal only with the sharpness of the font bends and not in which direction the bend curves.

As shown in FIG. 1, the process continues by calculating the curvature derivative at each resampling point 140. According to FIG. 7, the curvature derivative at R is defined as the absolute curvature at R + 1 minus the absolute curvature at R1 divided by two ( ie the slope of the curve of the plotted curves). Figure 9 graphically illustrates the curvature derivative obtained at each point shown in Figure 7. Similar to using more than two points to obtain a more accurate curvature value as described above, the curvature derivative could be calculated using a wider window (five points vs. three) where relevant points exist, and a narrower window (two points vs. three) there where necessary. Since the curvature derivative cannot be calculated at the endpoints of the written segment, the curvature derivative at the endpoints is simply defined as the same at the corresponding adjacent points.

Referring to Fig. 1a, a preferred embodiment of the method further examines a newly calculated array of curvature derivative values to determine points where the derivative is in its local ipaxime 910 (defined to include points at the end of the curve turnover and beginning to descend) or where the derivative is in its a local minimum of 920 (defined to include points at the end of the turnover beginning with ascending). For each entry area (in time) after the local maximum and before the local minimum, the center point of the section (expressed in section arc length) is determined to be '930'. This midpoint is defined - - as - M - 9-30 .- If the difference between the local maximum and local minimum values of the section exceeds the threshold value T 940 and the absolute value of the curvature at w M exceeds a certain threshold value T

Í820 ’;

selected as stroke boundary 150.

The parameters T and T must be estimated and depend on the units in which the curvature and the curvature derivative are measured. Exact values are not critical here when an error tolerant character recognizer is used. In performing any experimental tuning of these or any other parameters to / create a specific embodiment of the invention, the desired goal is to achieve as homogeneous segmentation as possible for many cases of individual character classes. This should be done empirically, by examining how the procedure segments the various actual font samples to be recognized.

JIn addition to the selected stroke boundary points described above, points where the pen is raised or lowered are selected as stroke boundary points. In a preferred embodiment of the present invention, the boundary points based on the curvature derivative may be shifted to. by two points where absolute values By shifting the boundary points, the determination of the points can be improved, causing the curvature of the established boundary to fall to a maximum of 160. to derive the curvature of stroke points by providing both the curvature rate and the curvature derivative provide the same stroke boundaries. However, this preferred application of the present invention is performed for a given point only if no stroke consisting of less than three points is created.

The set of boundary stroke points is defined in accordance with the present invention by a set of corresponding strokes. These strokes can then be passed to the character recognition based on the strokes to recognize handwriting input.

The present invention and its preferred embodiments relate to new, more accurate stroke segmentation methods. In accordance with the present invention, in many cases of handwritten input, the input is repeatedly divided into similar sets of strokes each time. For example, if the handwritten entry is an L written at different times by different writers, the present invention and its preferred embodiments would each time more precisely divide the L entry into similar stroke segmentation breakpoints regardless of the differences of the different writers. Such stroke segmentation would help to provide a more accurate interpretation by stroke-based recognizer.

Those skilled in the art will recognize that many embodiments of the present invention may be used. One obvious extension is from the case of the printed manuscript described herein to the case of the combined script. The present stroke segmentation method is independent of the character separation method, and therefore, the blended font techniques can easily utilize stroke segmentation. By another embodiment, the scanned or unbroken font would apply the present invention to a font thinning problem such that a constant curve is obtained. The stroke boundaries could then be determined allowing processing of the method described herein could be stroke segmentation. A straightforward method would consist of a width written at both the curvature derivative points and the intersection points, because the lack of time information makes the intersections and the touching bends look similar.

Claims (7)

1. A method of recognizing a handwritten character in a set and a set. colored pixels, comprising the steps of: calculating the value of the curvature derivative for each of the set of colored pixels, where each value of the curvature derivative represents the rate of change of the absolute curvature in the corresponding pixel, - selecting a set of stroke boundaries such that each stroke boundary lies between a colored pixel with a high curvature derivative value and a subsequent colored pixel with a low curvature derivative value, - specifying a set of strokes such that each stroke boundary is located at the end of the stroke, -i, calculating at least one stroke property value for each stroke to form a set of character properties, using the feature set to determine the identity of said handwriting. The method of claim 1, wherein the high-value curvature colored pixel has a locally maximum curvature derivative value and the subsequent low-curvature colored pixel has a locally minimum curvature derivative value. The method of claim 2, wherein each stroke boundary lies at a point polov half the distance between the colored pixel having a locally muximal curvature derivative value and the colored pixel having a locally minimal curvature derivative value. The method of claim 1, wherein each stroke boundary lies at a locally maximum absolute curvature point. A method of recognizing a handwriting feature composed of a sequence of points, wherein each point consists of ž values ^- three spatial coordinates, comprising the steps of: calculating the value of the curvature derivative. for each of the set of points, where each value of the curvature derivative represents the rate of change of the absolute curvature at the corresponding point, - selecting a set of stroke boundaries such that each stroke boundary lies between a point with a high curvature derivative value and a subsequent point with a low curvature derivative value, - specifying a set of strokes such that each stroke boundary is located at the end of the stroke, - calculating at least one stroke property value for each stroke to produce a set of character properties, using the feature set to determine the identity of said handwriting. The method of claim 5, wherein the high curvature derivative point has a locally maximum curvature derivative value and the subsequent low curvature derivative point has a locally minimal curvature derivative value. The method of claim 6, wherein each stroke boundary lies at a point half-distance between a point with a locally-maximum value of the curvature derivative and a point with a locally minimum value of the curvature derivative. The method of claim 5, wherein each stroke boundary lies at a locally maximum absolute curvature point. O 03 · í / 3 UD '<T> D031CI <>. $ OSR. . / . / \ • --Λ '. o -χο X \ \ • \ s Λ \ 'in Axis. . 4 400 · 3/3 6/0 o what LO C> 00, 10 h e 'ť O3Z 500 ... 9 • • ~ OSR. 6 ·· · » and. 7 7-77 Description of picture no 110 - receiving input as a sequence of (x, y) points 120 - resampling so that points are equally spaced, 130 - Calculate the amount of curvature at each point X · W »X i -4Q yý ^ count 'deriv ^ c' 'ts-k? / every point Λ .. / ..... --y - '7 150 - Identification of Stroke Limits as' points in the middle of the limits at the local maximum and minimum curvature derivative 160 - Adjustment of stroke limits to correct small measurement errors 170 - passing detected stroke boundaries for further use in the recognition process