CN1106548A

CN1106548A - Method and apparatus for inputing handwriting

Info

Publication number: CN1106548A
Application number: CN94109513A
Authority: CN
Inventors: 茶谷公之; 前地三千儿
Original assignee: Sony Corp
Current assignee: Sony Corp
Priority date: 1993-08-13
Filing date: 1994-08-03
Publication date: 1995-08-09
Anticipated expiration: 2014-08-03
Also published as: JPH0757054A; CN1095142C; JP3353956B2; KR950006649A; KR100312462B1

Abstract

The aim is to accurately recognize handwritten character. When characters are inputted, six subordinate characters (a)-(f) are generated from those characters. Then only a necessary number of adjacent characters among the six subordinate characters are combined. Further, the respective characters are recognized from the respective combinations.

Description

Method and apparatus for inputing handwriting

The present invention relates to for example be used for the splendid hand-written character input method and the device of devices such as notebook type electronic installation and portable microcomputer.

Recently, for example notebook type electronic installation etc. can be popularized with the device of handwriting mode input characters.This can be different with the device of handwriting mode input characters and common word processor etc., not with keyboard but with pen literal is just write on the panel can input characters, so, as literal being write on the paper with pencil etc., input characters very like a cork.With the situation of using keyboard relatively, needn't remember the position of key, so the beginner is input characters at an easy rate also.

This can have two kinds with the device of handwriting mode input characters, a kind of input frame of setting input characters, and another kind is not set input frame, can be at position input characters arbitrarily.

In the device of setting input frame, the user is input to literal in this input frame by 1 word ground of hand-written 1 word.This input frame is divided into a plurality of zones usually, is input to each zone 1 literal of 1 literal of user.Because the position of input is restricted, so, concerning the user, this restriction is made troubles to operation, still, because the zone of each literal of input separates, so, just cut open 1 literal of 1 literal of the literal that will not import and can be familiar with each literal more exactly.

In contrast, in not setting the device of input frame, the user is input to optional position on the input panel (flat board) to literal with handwriting mode.At this moment, for example discern like that by the literal of handwriting mode input according to flow process shown in Figure 48.

That is, during initial input characters, under the condition of appointment, discern, and cut out (segmentation of identification row) by the row that literal constituted of input.And, after the cutting trip, the literal in this row is cut open under the condition of appointment 1 literal of 1 literal.That is, determine the literal of input and the boundary of literal.At last, each literal of cutting open is discerned as 1 literal, and shown this recognition result.

As a result, for example, shown in Figure 49 (a) like that, when the user has imported " bright " this Chinese character, when " day " and " moon " word being cut into 1 literal of formation, just take the literal of importing as Chinese character " bright ".

But, shown in Figure 49 (b) like that, if write to such an extent that the interval between " day " and " moon " is bigger, respectively when independently literal is cut open, these two literal will be taken as independently literal respectively in " day " and " moon " word.

What at this moment, just can't judge that the user imports is " bright " word or " day " and " moon " word.

Like this, for the device of not setting input frame that has earlier with the handwriting mode input characters, with regard to exist one according to the cut state of literal the big problem of possibility of misidentification literal.

The present invention develops in view of the above problems, and purpose is to discern handwriting more accurately.

Hand-written character input method of the present invention, it is characterized in that: at the handwriting characteristic of each stroke of literal that detects handwriting input, have the regular and sub-character cutting of row cutting rule in characteristic quantity recognin character that utilization detects and the hand-written character input method with identification literal after the sub-character combination, when the input stroke, row according to the feature of this stroke, sub-character and the input of row cutting rule judgment, when the row of the row of input and input in the past is during with delegation, carry out the judgement of sub-character according to the feature of stroke, sub-character and the cutting of sub-character are regular.

Hand-written character input method of the present invention, it is characterized in that: at the handwriting characteristic of each stroke of literal that detects handwriting input, utilize in the characteristic quantity recognin character and hand-written character input method that detects identification literal after the sub-character combination, the row of the literal that identification is made of stroke, and when change literal capable, in the lump the laggard style of writing word of sub-character combination is discerned.

In addition, hand-written character input method of the present invention, it is characterized in that: at the handwriting characteristic of each stroke of literal that detects handwriting input, utilize in the characteristic quantity recognin character and hand-written character input method that detects identification literal after the sub-character combination, each input stroke, during the sub-character that carries out that the cutting of sub-character and cutting make new advances, one by one with the laggard style of writing word identification of sub-character combination, simultaneously, the row of the literal that identification is made of stroke, and when change literal capable with the identification of the laggard style of writing word of sub-character combination.

In the identification of this seed character or literal, can use the sum of the stroke of sub-character or literal at least.

In addition,, can set up database to a plurality of conditions in advance, calculate the sufficient degree of each condition respectively, and the summation and the reference value of the result of calculation of each condition compared in order to carry out the cutting of sub-character or row.

In hand-written character input method of the present invention, when the input stroke, row according to the feature of this stroke, sub-character and the input of row cutting rule judgment, and in the behavior of the row of input and input in the past during, carry out the judgement of sub-character according to the feature of stroke, sub-character and the cutting of sub-character are regular with delegation.Therefore, can cut sub-character exactly.

In hand-written character input method of the present invention, the row of identification literal and when changing this row in the lump with the identification of the laggard style of writing word of sub-character combination.Therefore, can discern literal exactly.

In addition, in hand-written character input method of the present invention, when the change row,, when the sub-character that cutting makes new advances, carry out the identification of literal simultaneously one by one with the laggard style of writing word identification of sub-character combination.Therefore, can be rapider and identification exactly.

Fig. 1 is the structured flowchart of an embodiment of using the device of hand-written character input method of the present invention.

Fig. 2 is the process flow diagram of the action of the embodiment in the key diagram 1.

Fig. 3 is that the row of S13 cuts the process flow diagram of the more detailed process of judgment processing in the key diagram 2.

Fig. 4 is that the sub-character of S15 in the key diagram 2 cuts the process flow diagram of the more detailed process of judgment processing.

Fig. 5 is the process flow diagram that the more detailed process of handling is discerned in the sub-character combination of S19 in the key diagram 2.

Fig. 6 is the figure of the object lesson of description line cutting.

Fig. 7 is the figure of the object lesson of the sub-character cutting of explanation.

Fig. 8 is the figure of the object lesson of the sub-character combination identification of explanation processing.

Fig. 9 is the figure of other object lesson of the sub-character combination identification of explanation processing.

Figure 10 is the process flow diagram of the 1st row cutting of explanation rule.

Figure 11 is the process flow diagram of the 2nd row cutting of explanation rule.

Figure 12 is the process flow diagram of the 3rd row cutting of explanation rule.

Figure 13 is the process flow diagram of the 4th row cutting of explanation rule.

Figure 14 is the process flow diagram of the 5th row cutting of explanation rule.

Figure 15 is the process flow diagram of the 6th row cutting of explanation rule.

Figure 16 is the process flow diagram of the 7th row cutting of explanation rule.

Figure 17 is the figure of the 1st row cutting of explanation rule.

Figure 18 is the figure of the 2nd row cutting of explanation rule.

Figure 19 is the figure of the 3rd row cutting of explanation rule.

Figure 20 is the figure of the 4th row cutting of explanation rule.

Figure 21 is the figure of the 5th row cutting of explanation rule.

Figure 22 is the figure of the 6th row cutting of explanation rule.

Figure 23 is the figure of the 7th row cutting of explanation rule.

Figure 24 is the process flow diagram of the 1st sub-character cutting rule of explanation.

Figure 25 is the process flow diagram of the 2nd sub-character cutting rule of explanation.

Figure 26 is the process flow diagram of the 3rd sub-character cutting rule of explanation.

Figure 27 is the process flow diagram of the 4th sub-character cutting rule of explanation.

Figure 28 is the process flow diagram of the 5th sub-character cutting rule of explanation.

Figure 29 is the process flow diagram of the sub-character cutting rule among explanation Figure 27.

Figure 30 is the process flow diagram of the sub-character cutting rule among explanation Figure 28.

Figure 31 is the process flow diagram as the 1st sub-character synthesis condition of literal synthesis condition of S66 in the key diagram 5.

Figure 32 is the process flow diagram as the 2nd sub-character synthesis condition of literal synthesis condition of S66 in the key diagram 5.

Figure 33 is the process flow diagram as the 3rd sub-character synthesis condition of literal synthesis condition of S66 in the key diagram 5.

Figure 34 is the figure of the form of explanation stroke feature data.

Figure 35 is the figure of the form of the single lteral data database of explanation.

Figure 36 is the figure of the form of description line data database.

Figure 37 is the figure of the form of description line data.

Figure 38 is the figure of the form of explanation recognition result data database.

Figure 39 is the figure of the form of explanation recognition result.

Figure 40 is the figure of the form of description line cutting rule database.

Figure 41 is the figure of description line cutting regular record form.

Figure 42 is the figure of the form of the sub-character cutting of explanation rule database.

Figure 43 is the figure of the sub-character cutting of explanation regular record form.

Figure 44 is the process flow diagram of other action example of the embodiment in the key diagram 1.

Figure 45 is the figure of the more detailed processing of S232 among explanation Figure 44.

Figure 46 is the figure of the processing of S231 among explanation Figure 44.

Figure 47 is the figure of the processing of S231 among explanation Figure 44.

Figure 48 is the process flow diagram that the literal identification treatment scheme that has earlier is described.

Figure 49 is the figure that the literal identification processing that has earlier is described.

1-CPU

2-ROM

3-RAM

The 8-bus

The 11-display controller

12-V-RAM

The 13-display

14-input panel interface

The 15-input panel

The 16-pen

Fig. 1 is the structured flowchart of an embodiment of using the device of hand-written character input method of the present invention.CPU1 carries out the literal input action according to ROM2 program stored control each several part.In RAM3, store data designated etc. as required.Standby RAM4 thinks canned data by the battery 5 frequent voltages of supplying with appointment so that also can store when the power supply of shearing device.Timer 6 is a benchmark with quartz crystal unit 7, keeps its frequent action by battery 5.

Display controller 11 will launch on V-RAM11 by the data of bus 8 inputs, and the image that launches is thereon shown on display 13.By operating pen 16, input panel 15 is defeated by panel interface 14 with the coordinate position data of input, and panel interface 14 is defeated by CPU1 or display controller 11 by bus 8 with these data.

Microphone 18 is defeated by solid recording treatment circuit 17 with the voice signal that receives, and stores.In addition, the voice signal of solid recording treatment circuit 17 outputs carries out playback by totalizer 20 supply loudspeakers 21.Dialer (dial) 19 is defeated by loudspeaker 21 by totalizer 20 with the dialing signal of appointment according to the control of CPU1.

When the action of switching device shifter, operate a plurality of key switch groups 22, the signal corresponding with this operation is defeated by CPU1 by bus 8 from keyed jointing mouth 23.In addition, by operating a plurality of key switch groups' 24 assigned switch, energized circuit 25, to the each several part power supply, can carry out solid recording treatment circuit 17 recording, playback, stop and action such as volume adjustment.

Can insert not shown integrated circuit card (IC-card) in the

socket

26,27 pairs of these IC-cards of IC-card interface carry out the conveying and the reception of data.

Plug pecker 29 can connect for example modulator-demodular unit etc., can carry out the conveying and the reception of signal by expansion interface 28 and modulator-demodular unit.

In addition, input panel 15 is made of transparent material, be formed on display 13 above.Like this, the demonstration of display 13 just can be seen by input panel 15.

Below, with reference to the action by the handwriting mode input characters in the device of the flowchart text said structure of Fig. 2.This processing is undertaken by CPU2 according to the ROM2 program stored.The user constitutes 1 stroke arbitrarily of literal with pen 16 each input on input panel 15, and CPU1 carries out the processing shown in the process flow diagram among Fig. 2.In this process flow diagram, the variable sum of expression stroke number represents that the variable S of the described sub-number of characters in back is initialized as 0 when energized.

In addition, carry out in this process flow diagram each and handle needed database storing in ROM2 or RAM3.

Beginning at step S11, is extracted the feature of input stroke.And this feature is defeated by stroke feature database D B1 stores.About being stored in the form of the data among this stroke feature database D B1, the back is introduced with reference to Figure 34.

Then, enter S12, judge whether the person's handwriting of input is initial person's handwriting after the line data initialization.That is, judgment variable i whether be 0(as described later, this variable i will increase by 1 at S14, so, when the 2nd stroke is imported later on, become value greater than 1).When the person's handwriting of input is person's handwriting initial after the line data initialization, enter S20, processing enters a new line.Generate new data line at this, and variable i and S are changed to 1 respectively.

At S12, after not being the line data initialization, the person's handwriting of concluding input during initial person's handwriting, then do not enter S13, the processing that the cutting of going is judged.In this is handled, use what S11 generated to be stored in stroke feature data among the stroke feature database D B1, to be stored in be expert at row cutting rule among the cutting rule database DB2 of line data among the DB4 of line data storehouse and storage.Judge whether the stroke of input now is included in the row of being represented by line data (be input in the identical row or be input in the different row) at this.About the details that this row cutting is judged, the back is introduced with reference to Fig. 3.

In the feature extraction data of up-to-date stroke, for example, except the coordinate of unique points such as the initial point of stroke or terminal point, also comprise and external rectangular data of stroke etc.In addition, about row cutting rule, to be introduced with reference to Figure 10～Figure 23 in the back, still, will use the projected length of coordinate, each stroke of bounding rectangle of coordinate, stroke number, each stroke of the bounding rectangle of the row that handwriting data forms and physical length etc. here.

In the cutting judgment processing of being expert at, as long as not concluding the new input stroke of this input is input to and the different row of importing so far of stroke, promptly conclude when being input in the row identical with the stroke of input so far, just enter S14, with the stroke feature data supplementing of the stroke of this input in the line data of line data storehouse DB4.And, make the variable i of the stroke number in the expression line data increase by 1.

Then, enter S15, carry out sub-character cutting judgment processing.Here, so-called sub-character is meant the set of the person's handwriting of the key element that constitutes 1 literal.Promptly, in the present embodiment, 1 literal is decomposed into the sub-character as key element, judge and respectively import the stroke that stroke is the formation sub-character identical with the stroke of previous input, still the stroke of the sub-character that the stroke of the stroke of the sub-character that formation is identical with a plurality of strokes of more than one preceding input or formation and input so far is different.

In the judgment processing of this sub-character cutting, use the line data of storing among the sub-character cutting rule of in sub-character cutting rule database DB3, storing and the database D B4 that is expert at.The details of this judgment processing, to be introduced with reference to Fig. 4 in the back, but, for example be to consider to judge behind the lap of level that the level of up-to-date stroke person's handwriting or vertical projection composition and already present sub-character script are gathered or vertical projection composition.

At S15, when the person's handwriting of up-to-date stroke is the person's handwriting of the formation sub-character identical with last sub-character (S-1 sub-character), enter S16, append processing.Here, up-to-date stroke data is appended in S-1 the sub-character data.That is, handle as the person's handwriting that constitutes the sub-character identical with last individual sub-character.Variable S keeps this value constant.

At S15, when the person's handwriting of concluding up-to-date stroke and previous more than a plurality of sub-character data become one, when constituting 1 sub-character, that is, conclude when constituting 1 sub-character with the n from S-n to S-1 sub-character, enter S17, carry out override and handle.That is, in the reason, will constitute 1 sub-character to the individual sub-character of the n of S-1 sub-character as person's handwriting from S-n the sub-character that has generated and handle, thereby reduce 1 sub-character herein with this input.And variable S is changed to S-n.

In addition,, when the person's handwriting of concluding up-to-date stroke constitutes different with the sub-character that generates so far new sub-character, just enter S18, carry out the fresh character processing at S15.That is, in the reason, up-to-date stroke data is handled as constituting new sub-character herein, and variable S increases by 1.

The stroke of new input be input to different so far row in before, carry out the processing shown in the above S11-S18 repeatedly.

And, when the stroke of concluding new input at S13 be input to different so far row in the time, just enter S19, be input to sub-character in the row up to now by combination, carry out the identification of literal and handle.In this identification is handled, use the line data (sub-character data) that is stored among the DB4 of line data storehouse.Be defeated by recognition result database D B5 according to the lteral data that recognition result obtains, store.

That is, in the present embodiment, being input as opportunity at new row, the sub-character that is input to the person's handwriting in the row up to now by combination begins to discern the processing of literal.

Then, enter S20, generate the new data corresponding, and variable i and S are changed to 1 with the stroke of this input.

Below, with reference to the process flow diagram among Fig. 3, the details in the capable row cutting judgment processing of S13 being described, beginning will be gone the counting 1r of cutting number of regulation at S31 and will be initialized as 0.Then, at S32, read the regular L_Rule[1r of row cutting that stores among the cutting rule database DB2 that is expert at].In addition, read line data from line data database D B4.

In this line data database D B4, storing sub-character data that the result according to S16～S18 among Fig. 2 obtains and the stroke number i that supplies with by S14.This line data is applied to the regular L_Rule[1e of row cutting], calculate the judgment value g[1r that is used to judge whether to go cutting (whether as the processing of importing to new row)], (g[1r]=L_Rule[1r]).The judgment value g[1r that calculates] be defeated by row cutting rating database DB12 and store.

Then, enter S33, judge whether to have read all row cutting rules (whether 1r=MAX_LR) from row cutting rule database DB2, and, when having the row cutting rule of not reading, just enter S34, make the counting 1r of row cutting sequence number increase by 1.Then, turn back to S32, carry out same processing repeatedly.That is,, can obtain and be expert at corresponding judgment value g[1r of a plurality of row cutting rules among the cutting rule database DB2 of all storages] by carrying out this processing repeatedly.

Then, enter S35, from row cutting rating database DB12, read the corresponding judgment value g[0 of each row cutting rule with above calculating]～g[MAX_LR] value, and synthesize (additive operation), to variable G assignment.

Then, enter S36, will compare at composite value G and the predefined reference value G_1mL that S35 calculates.When composite value G greater than as minimum reference value G_1mL the time, conclude that then the person's handwriting of this input is input in the row different with row up to now.In contrast, when concluding that composite value G is equal to or less than reference value G_1mL, the person's handwriting of then concluding this input is input in the row identical with up to now row.

Below, cut the details of judgment processing with reference to the sub-character of S15 in Fig. 4 key diagram 2.Beginning at S41, will represent that respectively the variable n of override number is initialized as 0, and will represent that the variable j of sub-number of characters counting is changed to S-1.Then, at S42, the variable Sr that sub-character cutting number of regulation is counted is initialized as 0.

Then, enter S43, read the sub-character that is stored among the sub-character cutting rule database DB3 and cut regular S_Rule[Sr].In addition, the data of up-to-date stroke in the line data that is stored in the DB4 of line data storehouse are applied to the sub-character of reading now cut regular S_Rule[sr] in, calculate the judgment value g[sr that is used to judge whether to carry out sub-character cutting] (g[sr]=S_Rule[sr]).And will obtain judgment value g[sr according to result of calculation] be defeated by sub-character cutting rating database DB22 and store.Cut regular S_Rule[sr about sub-character] details, will be introduced with reference to Figure 24～Figure 30 in the back.

At S44, the processing of S43 is stored in the sub-character cutting rule database DB3.Do not obtain and all corresponding judgment value g[sr of sub-character cutting rule when concluding] time, just enter S45, make variable sr increase by 1, carry out the later processing of S43 once more repeatedly.

Like this, when conclude obtained be stored in sub-character cutting rule database DB3 in the corresponding judgment value g[sr of all sub-characters cutting rules] time, just enter S46, from sub-character cutting rating database DB22, read this judgment value g[sr], and synthesize (additive operation) at S46.Then, give variable G with this composite value.

Then, at S47, will compare at composite value G and the predefined reference value G_1mS that S46 calculates.When concluding that this composite value G is equal to or less than minimum reference value G_1mS, the stroke of then concluding present input is the input to identical sub-character, then, just enters S48, and the variable n amount of representing the override number is variable S-j.Then, after S49 makes variable j subtract 1, enter S50.

At S50, judge whether the stroke of j sub-character [j] and input now satisfies sub-character synthesis condition.Perhaps whether judgment variable j is less than 0.That is, subtract 1 by making variable j, note last sub-character at S49, when this sub-character and input stroke satisfy sub-character synthesis condition or variable j greater than 0 the time, just turn back to S42, sr is changed to 0 with variable, carries out processing thereafter repeatedly.

That is,, sub-character cutting rule used in j all sub-character, generate judgment value g[sr] by carrying out above action repeatedly.

As the sub-character synthesis condition at S50, for example, whether the external quadrilateral that can be used as sub-character [j-1] and sub-character [j] has the condition of intersection point.At this moment, when not having intersection point, just turn back to S42.

On the other hand, at S47, when concluding composite value G, conclude that then the stroke of present input constitutes the sub-character different with sub-character so far, so the processing of skipping above-mentioned S48 is carried out S49, the processing that S50 is later repeatedly greater than minimum reference value G_1mS.That is, at this moment, make variable n keep value in the past constant.Like this, just, variable n is changed to the override number.

Then, at S50, when concluding that sub-character and input stroke satisfy sub-character synthesis condition or when concluding that variable j is negative value (j＜0), just enter S51, the value of judgment variable n is much.When variable n is 1, enter S16, append processing (1 override is handled); When variable n greater than 2 the time, enter S17, carry out override and handle; When variable n is 0, enter S18, carry out fresh character and handle.

Below, discern the details of handling with reference to the sub-character combination of S19 among the flowchart text Fig. 2 among Fig. 5.In the present embodiment, the K from i to j sub-character handled, beginning at S61, is initialized as 0 with the variable i that begins sub-character count in the process flow diagram, and the variable j that will finish sub-character count at S62 is changed to i.

Then, enter S63, K the sub-character that reads from line data database D B4 from i to j makes up, and generates 1 lteral data, and is defeated by single lteral data database D B31 and stores.

Then, carrying out literal identification at S64 handles.That is, read the single lteral data that is stored among the single lteral data database SB31, discern this literal, and will be defeated by recognition result data database DB5 according to the data that recognition result obtains and store.

Then, enter S65, make variable j increase by 1.That is, make the sub-character of end increase 1(and change to nearest sub-character).Then, enter S66, judge whether an i and j character satisfies the literal synthesis condition, and perhaps whether judgment variable j equals s.

S65 make finish sub-character and increase by 1 after, when concluding that i and j sub-character satisfy the literal synthesis condition, just turn back to S63, these two sub-characters are generated single lteral data once more, carry out literal identification repeatedly and handle.When variable j is not equal to s, promptly the same when the value j that finishes sub-character count is not equal to sub-number of characters s when also not reaching last 1 sub-character (finish sub-character) with above-mentioned situation, turn back to S63, carry out same processing repeatedly.Details about the literal synthesis condition will be introduced with reference to Figure 31～Figure 33 in the back.

At S66, when concluding that an i and j character does not satisfy the literal synthesis condition or when concluding that variable j equals s when being not equal to s sub-character (finish sub-character), enter S67, make variable i increase by 1.That is, only will begin sub-character changes to up-to-date.And when S68 concludes that variable i is not equal to s, turn back to S62, variable i is changed to variable j, carry out processing thereafter repeatedly.That is, change under the state of up-to-date sub-character only will beginning sub-character, carry out same processing repeatedly.

And, before S68 is concluding that the sub-character of beginning equals up-to-date sub-character, carry out same processing repeatedly.

Above, for the situation of input handwriting illustrated cut with the cutting of sub-character and the cutting of being expert at after the identification carried out from sub-character to literal handle, in order to sharpen understanding, below with reference to the object lesson of Fig. 6 and Fig. 7 description line cutting action and sub-character cutting action.

Fig. 6 shows the S13 among Fig. 2 and the object lesson of the capable cutting action among Fig. 3.In the present embodiment, write out on input panel 15 with pen 16 and occupy L1 to 4 of the L4 words of composing a piece of writing.Be expert at and write among the L1 "

ては " printed words, write the printed words of " row cuts out " among the L2 that is expert at.Be expert at and write the printed words of " explanation The " among the L3, write the printed words of " ま The " among the L4 that is expert at.

For example, from row L1 when shifting to capable L2, the person of being to use line of input L1 " ては " printed words after go the input of " OK " word of L2.If the last literal of row L1 ", " and the front writing position of literal " OK " word of capable L2 between distance greater than 3cm.At this moment, cut regular L_Rule[0 with reference to the described row of Figure 10 according to the back], then predicate the different new capable L2 of capable L1 that writes out with up to now.

In addition, go distance L between the terminal point of capable frame of the writing position of " saying " word of L3 and row L2 ₂₃Vertical width L than the capable frame of capable L2 _2HLong, the writing position of row L3 to the distance on the capable frame base of row L2 than capable frame height L _2H1/4 short, and the capable frame height L of row L2 _2HAnd width W _2HAll greater than 5mm, when satisfying this condition, according to the back with reference to the regular L_Rule[6 of the described row cutting of Figure 16], conclude that then " saying " word write in the capable L3 different with up to now capable L2.

The height L of the beginning literal " " of row L4 _4HThe height L of capable frame greater than row L3 _3H2/3, and the distance of the vertical direction mid point of the base of capable frame of row L3 and literal " " is greater than 3mm.At this moment, cut regular L_Rule[4 with reference to the described row of Figure 14 according to the back], then predicate input to new row.

Below, with reference to the object lesson of the sub-character cutting judgment processing among S15 and Fig. 4 in Fig. 7 key diagram 2.Now, suppose that the user wants to import hiragana literal " な ".Beginning.When importing the 1st stroke (Stk1), generate sub-character [0] accordingly with the input of the 1st stroke, and will represent that the variable S of sub-character is changed to 1.

Then, when input the 2nd stroke, owing to the input of the 1st stroke with the 2nd stroke intersects, so, according to the described sub-character cutting rule in back, be identified as same sub-character.Therefore, the variable n that represents the override number is changed to 1, appends processing.Whereby, the data supplementing of the 2nd stroke to by in the corresponding sub-character [0] that generates of the input of the 1st stroke, is generated 1 sub-character [0] by 2 strokes.

When input the 3rd stroke, conclude that then this input is the new sub-character different with sub-character up to now, and will represent that the variable n of override number is changed to 0, carry out new character and handle.Whereby, generate the sub-character [1] corresponding with the data of the 3rd stroke.And the variable s that will represent sub-number of characters is changed to 2.

When input the 4th stroke, conclude that then this sub-character and sub-character [0] constitute same sub-character, thereby generate sub-character [0].And, conclude that further this sub-character [0] and sub-character [1] constitute same sub-character, are changed to 2 with the variable n that represents the override number.Then, carry out override and handle, sub-character [0] and sub-character [1] is synthetic, form sub-character [0], and sub-number of characters s is changed to 1.

Below, with reference to the object lesson of the sub-character combination identification processing among the S19 in Fig. 8 and Fig. 9 key diagram 2 and Fig. 5.

Beginning, as shown in the drawing when importing the printed words of " bright Ru い " as shown in Figure 8, under the situation of present embodiment, these literal generate 6 sub-character a～f.And as shown in the figure, only with the synthetic necessary number of the sub-character group of adjacency among these 6 sub-character a～f.Then, according to each each literal of combination identification.

In the embodiment of Fig. 9, input " speech ", " river " and, " page or leaf " three Chinese characters, generate five sub-characters of a～e therewith accordingly.This a little character of appropriate combination can obtain literal identification candidate according to each combination.

Below, with reference to the process flow diagram of Figure 10～Figure 16 and Figure 17～Figure 23 corresponding, the regular L_Rule[i of row cutting that stores is described in the row cutting rule database DB2 of Fig. 3 with each process flow diagram].In these figure, each symbol is expressed as follows data respectively.

Pns: the initial point of up-to-date stroke.

(DB1 obtains by the stroke feature data database)

Ppe: the terminal point of the last stroke of up-to-date sub-character

(B4 obtains by the line data database D)

NBx: the bounding rectangle of up-to-date stroke

(the stroke feature data by stroke feature data database DB1 obtain)

PBx: the bounding rectangle of the last stroke of up-to-date sub-character

(line data by line data database D B4 obtains)

LBx: the bounding rectangle that surrounds whole person's handwritings of this row

(line data by line data database D B4 obtains) "

LEN_MN 5mm

LEN0 3cm

LEN1 1.5cm

LEN2 5mm

LEN3 1cm

LEN4 3mm

LEN5 5mm

Figure 10 represents the regular L_Rule[0 of row cutting].Beginning at S81, judges that whether distance between the terminal point Ppe of last stroke of the initial point Pns of up-to-date stroke and up-to-date sub-character is greater than LEN0(3cm).When the distance of two points during, enter S82, with judgment value g[0 greater than 3cm] be changed to maximal value MAX_GRADE; When the distance of two points during, enter S83, with judgment value g[0 less than 3cm] be changed to 0.Under this regular situation, as shown in figure 17, utilize its grade of distance decision between the terminal point Ppe of last stroke of the initial point Pns of up-to-date stroke and up-to-date sub-character.

Judgment value g[i] in the bigger expression of numerical value to satisfy the possibility of this condition (situation of present embodiment is to new row input) big.

Figure 11 represents the regular L_Rule[1 of row cutting].In this rule, at S91, as shown in figure 18, whether the distance of the initial point Pns that judges up-to-date stroke and the bounding rectangle LBx that surrounds the whole person's handwritings of this row is greater than 1.5cm.Both distances are during greater than 1.5cm, at S92 with judgment value g[1] be changed to maximal value; Both distances are during less than 1.5cm, at S93 with judgment value g[1] be changed to 0.

Figure 12 represents the regular L_Rule[2 of row cutting].At S101, as shown in figure 19, whether the height of the bounding rectangle LBx of these all person's handwritings of row of judgement encirclement and width be all greater than 5mm(LEN_MN).And when its length during greater than 5mm, and then whether the vertical direction of judging the bounding rectangle LBx that surrounds the whole person's handwritings of this row and the bounding rectangle NBx of up-to-date stroke at S102 is apart from greater than 5mm(LEN2).

When two conditions among S101 and the S102 all satisfy, at S103 with judgment value g[2] be changed to maximal value, when any one condition does not satisfy, just at S104 or S105 with judgment value g[2] be changed to 0.

Figure 13 represents the regular L_Rule[3 of row cutting].Beginning, at S111, as shown in figure 20, whether the height of the bounding rectangle LBx of whole person's handwritings of this row of judgement encirclement and width be all greater than 5mm.And,, further judge that at S112 whether the distance of terminal point Ppe of last stroke of the bounding rectangle NBx of up-to-date stroke and up-to-date sub-character is greater than 1cm(LEN3 when its length during greater than 5mm).

When two conditions among S111 and the S112 all satisfy, at S113 with judgment value g[3] be changed to maximal value, when any one condition does not satisfy, just at S114 or S115 with judgment value g[3] be changed to 0.

Figure 14 represents the regular L_Rule[4 of row cutting].Beginning, at S121, as shown in figure 21, whether the height of bounding rectangle NBx of judging up-to-date stroke is greater than 2/3 of the height of the bounding rectangle LBx of the whole person's handwritings that surround this row.When satisfying this condition, just enter S122, and then whether the distance of judge surrounding vertical direction between the center of bounding rectangle NBx of the bounding rectangle LBx of whole person's handwritings of this row and up-to-date stroke is greater than 3mm(LEN4).

When two conditions all satisfy, at S123 with judgment value g[4] be changed to maximal value, when any one condition does not satisfy, just at S124 or S115 with judgment value g[4] be changed to 0.

Figure 15 represents the regular L_Rule[5 of row cutting].In this rule, as shown in figure 22, judge 3 conditions at S131～S133.At S131, whether the height of the bounding rectangle LBx of whole person's handwritings of this row of judgement encirclement and width be all greater than 5mm(LEN_MN).At S132, whether the bounding rectangle NBx that judges up-to-date stroke is positioned at distance is surrounded the base of bounding rectangle LBx of whole person's handwritings of this row below 1/4.In addition, at S133, judge whether the bounding rectangle NBx of up-to-date stroke is positioned at the about 5mm(LEN5 in the bounding rectangle PBx left side of the last stroke of up-to-date sub-character) locate.

When these three conditions all satisfy, at S134 with judgment value g[5] be changed to maximal value, when any one condition does not satisfy, just at S135～S137 with judgment value g[5] be changed to 0.

Figure 16 represents the regular L_Rule[6 of row cutting].In this rule, as shown in figure 23, judge three conditions at S141～S143.At S141, whether the height of the bounding rectangle LBx of whole person's handwritings of this row of judgement encirclement and width be all greater than 5mm.At S142, whether the bounding rectangle NBx that judges up-to-date stroke is positioned at distance is surrounded the base of bounding rectangle LBx of whole person's handwritings of this row and locates below 1/4.At S143, the terminal point Ppe left side whether the bounding rectangle NBx that judges up-to-date stroke is positioned at the last stroke of up-to-date sub-character surrounds the height place of the bounding rectangle LBx of the whole person's handwritings of this row.

When three conditions of S141～S143 all satisfy, at S144 with judgment value g[6] be changed to maximal value, when any one condition does not satisfy, just at S145～S147 with judgment value g[6] be changed to 0.

With each the regular judgment value g[i that obtains like this] synthetic, and this composite value G and reference value G_1mL compared, composite value predicates and imports in the new row during greater than reference value, and this is the same with the above.

Below, illustrate that with reference to Figure 24～Figure 30 the character in the sub-character cutting rule database DB3 that is stored among Fig. 4 cuts regular S_Rule[i].

Figure 24 represents that sub-character cuts regular S_Rule[0].Beginning, overlapping in the horizontal projection of the S151 horizontal projection of judging up-to-date stroke and the sub-character of having deposited whether less than 1/2 of either party's horizontal projection, if less than, then enter S512, with judgment value g[0] be changed to maximal value; If be not less than, then enter S513, with judgment value g[0] be changed to 0.

Figure 25 represents that sub-character cuts regular S_Rule[1].In this rule,, judge whether up-to-date stroke is positioned at the sub-character left side at S514.And, when concluding that up-to-date stroke is not positioned at the sub-character left side, just at S155 with judgment value g[1] be changed to maximal value; Just enter S156, with judgment value g[1 when concluding when being positioned at the left side] be changed to 0.

Figure 26 represents that sub-character cuts regular S_Rule[2].In this rule, whether the stroke number summation of judging up-to-date stroke and sub-character at S157 greater than 30, greater than 30 o'clock, just at S158 with judgment value g[2] be changed to maximal value; Be equal to or less than at 30 o'clock, just enter S159, with judgment value g[2] be changed to 0.

Figure 27 represents that sub-character cuts regular S_Rule[3].In this rule, for example, as shown in figure 29, carry out the cutting of sub-character and judge.As shown in the drawing, make that a big side is PB_SIZE in the length of vertical and horizontal of bounding rectangle of last stroke, the initial point of the bounding rectangle of last stroke is P_SP, the transverse width of the bounding rectangle of up-to-date input stroke is B_W, the leftmost point of up-to-date input stroke is L_P, and vertical width of row frame is LH.

At S311, judge that whether PB_SIZE is less than LH/5.That is, whether the length PB_SIZE that judges a side big in the length of vertical and horizontal of bounding rectangle of last stroke is less than 1/5 of vertical width HL of row frame.

At S312, whether the transverse width B_W of bounding rectangle that judges up-to-date input stroke is greater than vertical width LH of row frame.And the x coordinate P_SPx of initial point P_SP of bounding rectangle that judges last stroke at S313 promptly, judges whether initial point P_SP is positioned at the right side of a L_P whether greater than the x coordinate L_Px of the some L_P of the up-to-date input stroke leftmost side.

When three conditions among S311～S313 all satisfy, enter S314, with judgment value g[3] be changed to maximal value; When any one condition does not satisfy, just enter S315, with judgment value g[3] be changed to 0.

Figure 28 represents that sub-character cuts regular S_Rule[4].In this rule, as shown in figure 30, the transverse width that makes up-to-date input stroke is BW, and vertical width of row frame is LH, the maximal value of the y coordinate of the bounding rectangle of up-to-date input stroke is MX_BY, and the minimum value of the y coordinate of row frame (bounding rectangle of row) is MN_LY.

At S321, whether the transverse width BW that judges up-to-date input stroke is greater than vertical width LH of 3 times of capable frames.At S322, whether the maximal value MX_BY of y coordinate of bounding rectangle that judges up-to-date input stroke is less than vertical width LH sum of y coordinate minimum M N_LY with the 1/3 capable frame of the bounding rectangle of row.

When two conditions among S321 and the S322 all satisfy, enter S323, with judgment value g[4] be changed to maximal value; When any one condition does not satisfy, just enter S324, with judgment value g[4] be changed to 0.

This sub-character cuts regular S_Rule[i] judgment value g[i] big more, the possibility that satisfies this condition is also big more.Its composite value G and reference value G_1mS are compared, during greater than reference value, just cut out sub-character at last.

The details of the S66 Chinese words synthesis condition judgment processing in Figure 31～Figure 33 presentation graphs 5.

In the embodiment of Figure 31, carry out judging according to distance whether the sub-character from sub-character [i] to sub-character [j] constitutes the processing of 1 literal.At S161, the variable t that represents the 1st the sub-character that quilt is noted is changed to i, will represent that at S162 therefore the variable S of the 2nd the sub-character that quilt is noted is changed to t+1(, s=i+1 during beginning).Then, at S163 character [t] is set at variables D to the distance (during beginning, being the distance between sub-character [i] and the sub-character [i+1]) between the sub-character [s].

At S164, judge that this distance D is whether greater than 2 times the vertical width of capable frame.When distance D is equal to or less than 2 times the vertical width of capable frame, conclude that then these two sub-characters should synthesize (constituting 1 literal).And, enter S165, whether judgment variable S equals variable j,, whether has become last sub-character [j] in the sub-character that judgement should be noted that is.When not reaching last sub-character, enter S166, make variable S increase the sub-character that 1(only notes the 2nd quilt and change to new), and then turn back to S163, carry out same processing repeatedly.

At S165, when concluding that sub-character that the 2nd quilt noted has reached last sub-character [j], just enter S167, whether judgment variable t equals variable j.That is, judge whether the sub-character [t] that the 1st quilt noted has reached last sub-character [j].When concluding the sub-character [j] that does not reach last, enter S168, make variable t increase by 1.That is, the sub-character that the 1st quilt noted changes to next sub-character.Then, turn back to S162, it is bigger by 1 than the variable t that sets at S168 to set variable S, and carries out the later processing of S163 repeatedly.

At S167, (when the sub-character noted of the 1st quilt reaches the sub-character of being noted at last) transfers to processing shown in Figure 32 when concluding that variable t equals variable j.

At S164, when concluding distance D, conclude that then the condition that sub-character is synthesized 1 literal is false greater than vertical width of 2 times capable frame.

Like this, in the processing of Figure 31, judge distance between sub-character [i] and sub-character [j],, conclude that then the condition that sub-character is synthesized 1 literal sets up, thereby transfer to processing shown in Figure 32 if be not more than whether greater than vertical width of 2 times capable frame.

In Figure 32, beginning, the bounding rectangle that will comprise all bounding rectangles of sub-character [i]～sub-character [j] at S181 is set at Bx.At S182, the transverse width of the bounding rectangle Bx that will set at S181 is set at Bx _w, and be Bx at S183 with vertical width setup of bounding rectangle Bx _H

At S184, judge transverse width Bx in S182 and S183 setting _wWith vertical width B x _HRatio Bx _w/ Bx _HWhether less than 3.When this ratio less than 3 the time, conclude that then sub-character satisfies synthesis condition; When this ratio greater than 3 the time, then conclude and do not satisfy synthesis condition.

Like this, judge the wide Bx of the bounding rectangle of the whole person's handwritings that surround sub-character [i]～sub-character [j] exactly at S181～S184 _wWith high Bx _HRatio whether less than 3, when less than 3 the time, just carry out judgement shown in Figure 33.

In Figure 33, beginning is calculated the stroke number sum of sub-character [i] to sub-character [j] at S191, and this value is set at Nk.Then, whether judge this value Nx greater than 30,, conclude that then sub-character satisfies synthesis condition when less than 30 the time at S192; When greater than 30 the time, then conclude and do not satisfy the condition that sub-character is synthetic.

Like this, be exactly to judge that whether the stroke number summation of sub-character [i]～sub-character [j] is greater than 30 in processing shown in Figure 33.

When concluding Figure 31～when three conditions shown in Figure 33 all satisfy, last then conclude that the condition that sub-character is synthesized 1 literal sets up, when concluding that any one condition does not satisfy, conclude that then the condition that sub-character is synthesized 1 literal is false.

Like this, when the condition that sub-character is synthetic is set up (true time), just the S66 from Fig. 5 turns back to S63; (during puppet) just enters S67 from S66 when concluding that the condition that sub-character is synthetic does not satisfy.

Below, with reference to Figure 34～Figure 43 form in each database is described.

Figure 34 represents the embodiment of the form among the stroke feature database D B1.In this embodiment, stroke feature grade vector go～gn is configured in its front end.This stroke feature grade vector gi represent with 1 stroke shape figure of pre-prepd this stroke (T1 ..., degree of approximation Tn).

Unique point coordinate Co～Cm is distributed in thereafter, and these unique point coordinates Ci represents initial point, terminal point; The coordinate of mid point or salient point etc.

Then, that arrangement is the initial point of existing stroke and next stroke and the orientation dir_hh of terminal point～dir_tt.Dir_hh represents the orientation from the initial point of existing stroke to the initial point of next stroke.Dir_ht represents the orientation from the initial point of existing stroke to the terminal point of next stroke.Dir_th represents the initial point orientation from the terminal point of existing stroke to next stroke.In addition, dir_tt represents the orientation from the terminal point of existing stroke to the terminal point of next stroke.

Then, distribution be the initial point of existing stroke and next stroke and terminal point apart from len_hh～len_tt.Len_hh represents the distance from the initial point of existing stroke to the initial point of next stroke, and len_ht represents the distance from the initial point of existing stroke to the terminal point of next stroke.Len_th represents the distance from the terminal point of existing stroke to the initial point of next stroke, and len_tt represents the distance from the terminal point of existing stroke to the terminal point of next stroke.

Again, distribution be the apex coordinate KBx1 of bounding rectangle KB, KBY1, KBx2, KBY2.

Then, distribution be number of elements en, be element string eo～e with that _En-1Element ei makes person's handwriting be similar to the data of straight line, is made of the sign i of the lifting of initial point coordinate (xi, yi), orientation Di, length of element Li and expression pen.

Figure 35 is illustrated in the form of the single lteral data of storing among the single lteral data database D B31.As shown in the figure, that at first arrange is the apex coordinate SBX1 of the bounding rectangle of single literal, SBY1, SBX2, SBY2.That distribute then is the stroke number Kn of this single literal, then is the total length K1 of each stroke of this single literal.At last, distribution be stroke feature serial data Ko, K1 ..., K _Kn-1

Figure 36 represents the form of line data storehouse DB4.As shown in the figure, line data is arranged in order, has end of file symbol at last.

Figure 37 represents the form of each line data among Figure 36.Beginning, arrangement be the Lid of expression row ID, be stroke number Lnk then.Then, arrange sub-number of characters Lns, arrange the apex coordinate LBX1 of bounding rectangle then, LBY1, LBX2, LBY2.That arrange at last is sub-character data string S0, S1 ..., S _Lns-1

Figure 38 represents the form of recognition result database D B5.As shown in the figure, in this database, the recognition result data are arranged in order, and are last, have end of file symbol.

And the format structure of each recognition result data is shown in Figure 39.As shown in the figure, the sub-character ordinal number StSC of beginning is arranged in beginning earlier, arranges then and finishes sub-character ordinal number EdSc.This begins which sub-character that sub-character ordinal number represents that the initial sub-character of the substring of synthetic identification literal is this row, finishes sub-character ordinal number which sub-character that to represent to constitute last 1 sub-character of the substring of identification literal be this row.

Then, arrangement be that the literal candidate counts rn, arrange literal candidate symbol string R0 at last, R1 ..., R _Rn-1This literal candidate accords with Ri and is made of recognition result character code Ci and identification grade gi.

Figure 40 represents the form of row cutting rule database DB2.As shown in the figure, beginning, first queueing discipline is counted MAX_LR, is queueing discipline L_Rule[0 in order then], L_Rule[1] ..., L_Rule[MAX_LR_1].

The regular L_Rule[i of among Figure 40 each] form be shown in Figure 41.That is, instruction code i is arranged in beginning earlier, and operand ni is last then, series arrangement operation 0～operation ni-1.

Figure 42 represents the form of sub-character cutting rule database DB3.Beginning, first queueing discipline is counted MAX_SR, then, arranges each regular S_Rule[0 in order], S_Rule[1] ..., S_Rule[MAX_SR-1].

In addition, each regular S_Rule[i] record format as shown in figure 43.Instruction code i is arranged in beginning earlier, is operand mi then, and is last, series arrangement operation 0～operation mi-1.

In the embodiment of Fig. 2, conclude at S13 just to enter S19 when carrying out that the identification of the laggard style of writing word of sub-character combination that generates is so far handled, and still, this literal identification is handled also and can be carried out after 1 stroke input whenever carrying out to different row input.Process flow diagram when Figure 44 represents to handle like this.

The processing of S211～S220, the same with the processing of S11～S20 among Fig. 2.In the present embodiment, except these are handled, S17 and the S18 in S217 and S218(Fig. 2 respectively) also added the processing of S231 and S232 afterwards.

That is, in the present embodiment, imported new sub-character and when S218 concludes that carrying out fresh character handles, just entered S232, carried out sub-character combination identification and handle when concluding at S215.

In S232, as the sub-character of identifying object, be from sub-character [0] among S218 as new sub-character and the S-1 of additional sub-character [S-1] sub-character [S-2] before sub-character.

In contrast, as the sub-character of the identifying object that carries out at S219, because the identification here is the identification of carrying out after importing to new row, so, be S sub-character from sub-character [0] to sub-character [S-1].

When Figure 45 shows S232(S219 too) the more detailed processing example handled of sub-character combination identification.Its basic processing is identical with situation shown in Figure 5, still, in Figure 45, during beginning, at S261 variable i is set at es.This es represents the maximal value as the sub-character group's of identifying object footnote.That is the up-to-date sub-character of sub-character [es] expression.

Then, enter S262, j is set at variable i with variable, and enters S263.At S263, the sub-character from sub-character [i] to sub-character [j] is synthesized, generate 1 lteral data, and 1 lteral data that will generate is imported single lteral data storehouse DB31 and is stored.

At S264, read the single lteral data that is stored among the single lteral data storehouse DB31, carry out literal identification and handle, and the literal that will obtain according to the result of literal identification is defeated by recognition result data database DB5 and is stored.

Then, enter S265, just variable i reduces 1.That is, identifying object is transferred on previous (old) sub-character from up-to-date sub-character.And, enter S266, judge whether sub-character [i] and sub-character [j] satisfy the literal synthesis condition or whether judgment variable i equates with ss.Ss represents the minimum value as the sub-character group's of identifying object footnote.

At S266, when concluding that sub-character [i] and sub-character [j] satisfy the literal synthesis condition, just turn back to S263, sub-character [i]～sub-character [j] is synthetic, generate the processing of 1 lteral data repeatedly.Too, turn back to S263 when being not equal to the earliest sub-character [ss], carry out processing thereafter repeatedly as the sub-character [i] of identifying object.

When concluding that sub-character [i] and sub-character [j] do not satisfy the literal synthesis condition or as the sub-character [i] of identifying object, equaling the earliest sub-character [ss], then enter next step (under the S232 situation, then finish, under the situation of S219, then enter S220).

That is, in the embodiment shown in the figure, discern processing from old order, in the embodiment of Figure 45, then handle from new order to old to new according to sub-character [i].

Certainly, the situation of the S219 also situation with S232 is the same, also can be according to discerning processing from the up-to-date order to old.Perhaps conversely, though which kind of situation, can be according to discerning processing from old order to new.

On the other hand, the S215 in Figure 44 when the stroke of concluding present input should make up with a plurality of sub-character that has generated, just enters S217, carries out override and handles.And enter S231, make and finish sub-character for the recognition result when (at the sub-character of S217 from the sub-character before being set at S=S-n [S-(n-1)] to sub-character [s]) is invalid from sub-character [S+1] to the sub-character sub-character [S+n].That is, because these recognition results are summed up as 1 sub-character by carrying out the override processing, so make it invalid.

For example, as shown in figure 46, be provided with sub-character A now, B, C.At this moment, as recognition result, recognition result, the antithetical phrase character A of antithetical phrase character A arranged, the recognition result of the recognition result of B, antithetical phrase character B, antithetical phrase character A, B, the recognition result of C, antithetical phrase character B, the recognition result of the recognition result of C and antithetical phrase character C.

Under this state, import new stroke and conclude that at S215 it is that override is counted n at 2 o'clock, because n is the number of prepreerence existing sub-character, so, utilize sub-character B, the stroke of C and new input generates the new sub-character B of sub-character D().At this moment, do not need and sub-character B and the corresponding recognition result of C.That is, as shown in figure 47, because present situation is to have (being left) sub-character A and sub-character D, so, as recognition result, there are the recognition result of antithetical phrase character D, the recognition result of antithetical phrase character A, D and the recognition result of antithetical phrase character A.Therefore, the recognition result recognition result (in Figure 46) in addition that does not need antithetical phrase character A shown in Figure 46 at the recognition result shown in the scope E.So these recognition results are invalid.

As shown in figure 44, carry out new input at every turn, handle if carry out the identification of sub-character, then embodiment as shown in Figure 2 is such, and whenever the situation that just begins to discern processing when new row is imported is compared, can promptly finish identification and handle.

According to hand-written character input method of the present invention, when the input stroke, just cut the row of rule judgment input according to the feature of this stroke, sub-character and row, when the row of input with the behavior of importing in the past during with delegation, just according to the feature of stroke, sub-character and sub-character cutting rule, carry out the judgement of sub-character, so, can correctly cut sub-character.

According to hand-written character input method of the present invention, when change literal capable, be in the lump sub-character combination to be carried out literal identification, so, can correctly carry out literal and discern.

In addition, according to hand-written character input method of the present invention, with the behavior opportunity of change literal, when not only carrying out the identification of literal but also cutting new sub-character, be one by one the laggard style of writing word of sub-character combination to be discerned, so, can more promptly discern.

Claims

1, a kind of hand-written character input method, it is characterized in that: at the handwriting characteristic of each stroke of literal that detects handwriting input, utilize in the above-mentioned characteristic quantity recognin character and hand-written character input method that detects identification literal after the above-mentioned sub-character combination, have row cutting rule and sub-character cutting rule, when the above-mentioned stroke of input, feature according to this stroke, the row of above-mentioned sub-character and the input of above line cutting rule judgment, when the above line of input and the row of importing in the past are during with delegation, feature according to above-mentioned stroke, above-mentioned sub-character and above-mentioned sub-character cutting rule are carried out the judgement of above-mentioned sub-character.

2, a kind of hand-written character input method, it is characterized in that: at the handwriting characteristic of each stroke of literal that detects handwriting input, utilize in the above-mentioned characteristic quantity recognin character and hand-written character input method that detects identification literal after the above-mentioned sub-character combination, the row of the above-mentioned literal that identification is made of above-mentioned stroke, when when change row of above-mentioned literal, in the lump with the identification of the laggard style of writing word of above-mentioned sub-character combination.

3, a kind of hand-written character input method, it is characterized in that: at the handwriting characteristic of each stroke of literal that detects handwriting input, utilize in the above-mentioned characteristic quantity recognin character and hand-written inputting method that detects identification literal after the above-mentioned sub-character combination, the above-mentioned stroke of each input, all carry out the cutting of above-mentioned sub-character, when above-mentioned sub-character that cutting makes new advances, one by one with the identification of the laggard style of writing word of above-mentioned sub-character combination, simultaneously, the row of the above-mentioned literal that identification is made of above-mentioned stroke, when the row of above-mentioned literal changes, with the identification of the laggard style of writing word of above-mentioned sub-character combination.

4, by the described hand-written character input method of claim 1,2 or 3, it is characterized in that: in the identification of above-mentioned sub-character or literal, will use the sum of the above-mentioned stroke of above-mentioned sub-character or literal at least.

5, press the described hand-written character input method of arbitrary claim of claim 1～4, it is characterized in that: in order to carry out the cutting of above-mentioned sub-character or row, in advance a plurality of conditions are set up database, calculate separately the sufficient degree of each condition, and the summation and the benchmark of the result of calculation of each condition compared.

6, a kind of handwriting character inputting device is characterized in that: the described method of arbitrary claim of using claim 1～5.