WO2002071320A1

WO2002071320A1 - Inputting text into a hand-held device

Info

Publication number: WO2002071320A1
Application number: PCT/DE2002/000323
Authority: WO
Inventors: Markus Andert; Alexander Jarczyk; Hans RÖTTGER
Original assignee: Siemens Aktiengesellschaft
Priority date: 2001-03-07
Filing date: 2002-01-30
Publication date: 2002-09-12
Also published as: DE10110980A1

Abstract

The invention relates to a handwriting recognition method and device, whereby the symbol is determined by patterns of strokes and by one or more transition graphs. The inventive method involves: a first step, in which a first stroke is analyzed in order to determine its pattern; a second step, in which a transition graph is determined that has the pattern determined in the first step; a third step, in which additional strokes are analyzed for the same duration as the patterns of these strokes in the examined transition graphs, and; a fourth step, in which the symbol is determined that is assigned to the transition graph determined in the third step, said transition graph preferably being completely described by the input strokes.

Description

description

Text entry into a handheld device

The invention relates to a device and a method for handwriting recognition.

When entering text into a handheld device such as For example, in a PDA (Personal Digital Assistant) or pen cell phone, text recognition is performed by analyzing strokes. A stroke is a line or line on an input element of a hand-held device. The analysis is carried out either at the optical level, such as e.g. B. by optical cameras, or by analyzing capacity fluctuations on a surface. Various types of strokes can be entered via an optical camera. A letter, like the small "a *", is drawn in a continuous line, this type of entry is called Unistroke. If almost all letters and numbers are entered via such unistrokes, handwriting recognition can reach high values (> 95%). This method is used for recognition in some PDAs.

However, the user must choose between the input fields "Letters * and" Numbers *. For example, a vertical line from top to bottom in the letter field is an "i ^λλ , in the number field a" 1 *. In addition, upper and lower case is introduced upwards with a vertical line and corresponds to pressing the Shift key on a keyboard once. With this type of handwriting recognition, the user learns the syntax of the device. The device is unable to learn the user's handwriting. The user must therefore memorize a device-specific alphabet, which contains characters that sometimes only partially correspond to the natural spelling.

3 Z tr P- ö iQ <ö Φ Φ ι_ι. td tu tr CΛ ⁱ ^ l 3 3 ¹ ^ CΛ ω W Φ ^ PL ts ra CΛ Ö w- Φ CD Φ J Φ o PJ Φ F- H Φ Φ 0 Φ rt O 3 J F- Φ rt F- Φ 3 Φ PJ Φ H F- Ω 3 ri- F- to 3 ω ⁽ - ^■ H to n 3 H 3 Ω F ¹ 3 NH rt Ω 3 ω H φ H rt cn 3 H - Ω tr φ CΛ rt O: H φ Φ o E "Φ rt φ O iQ Φ tr W Ω O 3 ^{^} Φ cn φ Φ tr H P-

F- n- F- Φ Φ to F- <3 H- Ω tr CΛ H F- F- Φ H cn tr Λ "rt F- rt H IS 3 φ φ

P 3 F- H rt Ω o 3 <Q Ω tr PJ rt P- iQ Ω Φ 3 cn n H Φ Ω Φ O PJ F- 3 F- F- 3

CD Φ § 3 Hi • tr HH Φ ^{^} n PJ F- Φ tr 1 PL rt Φ tr F- PJ pr f H 3 3 3 tr

0 H- rt F- rt t- ¹ H tr PL. tr 3 3 t Φ H Ω φ φ 3 PJ rt 3 ^J tr ιP 3 cn Φ tu α 3 CΛ 3 0 F- tr F- J F- φ 13 CΛ Φ o 3 F- 3 ^J s: tr Hi 0 Φ Φ F ¹ P ⁾ iQ tr 3 φ

Φ F- * ϊa 3 Φ Φ 3 H Φ 03 3 to Ω rt Ω rt φ Φ P> Hi F- H F- tr Φ 3

1-1 cn 3 0 iQ iQ iQ iQ. Φ tr O "cn tr cn iQ 3 3 3 F- 3 φ tr to ι 0

5ö rt tr 3 Φ H Φ α F- N F- H Ξ F- O Φ rt ^ 2 F- ^ ιQ PL J O φ rt

<! φ o iQ 3 3 Φ σ J • ω F- φ 3 3 P φ iQ 0 Φ ⁾ cn 3 F "N

Φ F- F "ω F- P 3 tr ¹ F- Hi ω rt H) H PJ F ¹ Φ 3 tr Φ n 3 3 3 F- φ Φ n pf s: iQ rt φ N Φ φ 3: • rt P - rt φ ω ω E * J φ Ω H Ω PJ: cn φ Φ IQ H tr Φ Φ PL Φ rt F- to H Φ Φ φ Φ F- 3 N 3 3 £. r pf N - HH Φ

3 F- 0 3 P- <U Φ ω F- 3: F- H 3 o 3 3 3 Φ Φ S. 3 H 3 Φ Φ 3 3

3 <i ω H PJ ^' - Φ ü 3 r l- ¹ 3 3 tr 3 π • 3 cn PJ Hi PL ^ H F- 3 Φ Φ 3 Φ 3 o rr Ω So 3 o 3 F- O φ φ Φ Φ Φ O Ω 0- PJ J rt F ¹ 3 3 ^{^} tr 3

Φ 3 - Φ tr CΛ 3 Φ rt 3 H 3 3 F ¹ tr 3 tr. ω F ¹ PJ PL H Φ

3 3 g 1— 'Ό iQ iQ Z Jϊd 3 Ω H 3 ιQ PJ F- cn $ Φ Hi F- HT Φ

W φ S <Φ φ φ to rt Φ W e Φ J "F ¹ φ φ S tr 3 cn Hi HO rt F- to (- ^■ 3- Φ H 3 F- J 3 PJ PJ tr H F- Φ F- 3 PL to F- 3 Φ PJ Φ Φ F ¹ cn 3

F- Φ H cn H rüT Ω 3 Φ PL 3 ω F- to 3 Φ Φ H rt P- 3 H 3 iQ Φ Φ

3 3 φ rt Hi 3 Z Hi F- F- F ¹ I ¹ Hi? 3 rt C n Φ F- iQ Φ H Φ φ Φ

"* • 4 PJ PJ F- Φ 3 3 iQ ιQ • -3 O: 0- cn to tsi F- H Φ tö CΛ 3 PJ-- • - {Φ

3 φ H pf F ¹ HHJ Φ Φ Φ t- 0 3 ≥! HOJ rt N Ω α Φ r PL rt F- rt H Φ ιQ tr ¹ HH PJ 3 3 o O tr. r to ω φ s: H Φ φ M 3 z Φ ω <! Φ 3 H m ΪÖ PJ: F- ιQ Φ 3 Hl? R CΛ 3 ω Ω rt cn Φ O to 3 F- Hl o 3 Ω O 3 p- O: φ φ Φ rt 3 3 ω Hi PJ Ω 3. P- pf F- iQ K _"* 3 PJ tT 3 ^* 3 PL 3 & 3 F- Ω Φ F- φ 3 Φ 3 tr α F- J Hι 3 3 Φ d iQ Ω φ PJ t φ H Φ tr rt 3 ^J 3 3 o P- 3 3 3 F "α φ H rt φ 3 cn H PJ tr. 3 F- CΛ 0 H Ω PJ M Φ "3 3 F- to rjü φ F- Φ Hi? 3 ιp O F- tr Φ

Hl rt rt H pf α tr 3 Φ φ Φ cn Φ F- H CΛ PJ Φ N tr φ Φ is L - rt H 3 F- Φ PJ F "F- <! ≥5 to 1 3 φ ÖD Φ rt cn 3 3 0 F 'Φ

0 Φ o M 3 Φ Φ 3 3 Φ 3 o ^ Φ W 3 P - 1-1 Φ Φ F- rt Φ Φ cn Φ PJ N

H P. J PJ F- to F- ω 3 1 ω o rt F- 3 CΛ H H) Φ H CΛ H PJ 3 3

F- PJ φ 3 ψ 3 tr 3 Φ PL α ω tr CΛ N φ CΛ rt cn rt PJ rt 3 Hi 3

^ Φ 3 ω P pf H 0- Φ r φ 3 rt nrj O ¹ Φ F- rt Φ F- rt cn F ¹ H CΛ 3 Φ &

Hi ω PJ: <! H 3 0 Λ Φ HH F- H 3 ₍ _ ,. 3 Hi Φ • Φ 3 O rt φ s! Φ H

PL • 3 Ω 3 φ cn Hi 3 o 0 o O ω Φ Hl rt F- 3 3 ΪT F- 3 φ F- 3 3

F- 3 tr iQ li Φ φ ^ iQ 3 H? tr rt? to rt s: to H rt l Φ Hi iQ H 3 O: PJ =

Φ 1-1 w H F- Hi H 3 H o Φ F ¹ H PJ Φ PJ * Φ N? T cn rt Φ α Φ iQ pr

Ω F- F- ιQ J PJ: "« _* O 3 Φ 3 ω Φ o r__ HJ F- F- Φ DJ M • ω to Φ HF ¹ φ

« ^* 3 s Hi φ f rt NOH PJ 3 T 3 tr W iQ r F ^J 3 3 Ό Φ 3 F- n

PJ tr r 3 H Φ J φ Ω Φ ¹ P> Φ Φ Φ NN 3 rt σ F- rt O Ω

3 K et Φ Φ Φ • 3 to ^* Φ 3 3 to N 3 F, Φ Φ g Hi F- r N tu * ϋ tf PJ rt PJ tr HH 3 to K 3 Hi F- ^ 3 φ 3 Φ F- 3 Φ N rt 3 rt Φ 3

Φ 3 Φ iQ pr Φ O 3 P »Ω H Φ Ω 3 3 3 H cn Φ Ω F- 3

3 rt- HJ φ ω 3 HH F- 3 52! INI tr _^ - _^ . Φ WH tr DJ Φ 3 PL 3 3 "CΛ" * • 3

Φ ⁱ Q 3 3 ^• ö 3 Φ Φ Ω Φ F- rt ≥i H Φ rt Ω H PJ: 3 cn <ω to Ω P> tr 3 D> ιQ 3 1-1 PL H 3 3- ^ rt Hi « ^ 3 3 Φ tr Ω 3 3 OO rt tr Φ rt

Φ 3 to 0 0- ω "« rt ra N Hi F "1 Φ 3 3 3 tr F- ιQ d 3 J φ H 0- cn <Q iQ 3 Ω φ Φ Φ Φ O: CΛ 3 φ HO Cn cn Ω Φ H F- tr 3 i H rt MH iQ pf F- F- PL 3 H to rt 3 3 CΛ

! _ .. tr ^ rt e 3 o M cn Φ F- F ¹

> QP ⁾ φ 3 rt F- F- 3 Φ H Φ • Φ r Φ Φ rt tr 3 rt 3 tr F-

3 H Hi Φ to Φ to 3 O 3 F- 1-1 3 3 F- F- PL rt Ω

3 PJ Φ tr _^ rt 3 • ^* α H 3 F- 3 Φ Φ PL tr rt Hi 3 Φ PJ 3 3 Φ φ 3 Ω 3 F- Φ Φ

F- 3 3 ot pf Φ F- 3 H

which element follows another element. In a first step, a first stroke is analyzed to determine its pattern.

In a second step, a transition graph is determined based on the pattern recognized in the first step, which has the pattern determined in the first step.

In a third step, further strokes are analyzed as long as the patterns of these strokes are contained in the transition graph.

After the transition graph has been completely covered by the strokes, in the fourth step the symbol is selected which is assigned to the transition graph determined in the third step.

In exceptional cases it can happen that a transition count is examined for a long time for further strokes. However, the last stroke entered was none

If the transition graph matches, a return must be made until a transition graph is found that contains all strokes. The following step describes this situation. If a stroke is recognized, the pattern of which is not contained in the examined transition graph, the process jumps back to the previously recognized pattern until a transition graph is found which is completely covered by the patterns recognized up to this pattern, and then the corresponding symbol to determine. So that the strokes entered up to this point are not lost, the skipped strokes are used to analyze a new transition graph.

The further independent method claim shows a detailed alternative sequence of the method as described below in the corresponding figure. The symbols are also described by patterns and transition graphs. In a first step, a stroke is read and assigned to a pattern.

In a second step it is checked whether the pattern is a new entry, i.e. is the beginning of the entry of a new symbol, or a continuation of a previous entry.

In the third step it is checked whether the analyzed pattern represents a start pattern of a transition graph. In the event that it is not a start pattern, an attempt is made to determine a symbol for the pattern. If successful, it was a Unistroke. If a Unistroke could not be determined, an error message is issued.

In the event that the input is continued, an attempt is made to determine a symbol on the basis of the previously recognized transition graph. Here is the

The next step is to compare the pattern with the requirements of the transition graph.

In the event that there is another pattern in the transition graph or a starting pattern in the transition graph and the transition graph has no further patterns, the symbol is determined.

In the event that there is a further pattern in the transition graph or a start pattern in the transition graph and the transition graph has further patterns, proceed to step one in order to determine new patterns of the transition graph.

A status determines whether it is a new entry or a continuation of the entry, the status being set to start whenever a symbol has been determined or an unsuccessful attempt was made to determine the symbol.

As a rule, the symbol assigned to the transition graph is determined and output on a screen.

In a further advantageous embodiment, the recognized symbols are only output on the display when a word has been completely ended, the strokes previously entered being graphically represented in such a way that the recognized word can be derived from them. In order to speed up the input, word suggestions are made before the complete input is complete and can be selected by the user.

The actual dynamics of this process is through the

Learning ability of the system achieved. In a configuration step, preferably depending on a user, the transition graph, the patterns and their sequence can be dynamically assigned to each symbol or a series of symbols.

In order to ensure a sensible processing of the entered strokes, these must be entered within a prescribed time window so that they can be assigned to a symbol or its transition graph. If strokes are entered outside of this time window, they are assigned to a new symbol and a new transition graph must be determined. Another possibility, which makes it possible to determine which strokes belong to a symbol, consists in a corresponding command, which is determined either by a special key or by a special stroke.

The transition graphs are, preferably directed, counts, which determine the sequence of the entered patterns or their strokes. Another essential component of the present invention are devices that are controlled by the method described above. This is preferably a mobile input device for handwriting recognition of symbols, a symbol being determined by patterns of strokes and / or a plurality of transition graphs. The input device has an element for recognizing strokes and a processing unit which carries out the analysis of the pattern according to one of the methods described above.

A pen, which is moved on the surface of the input element, is preferably used to enter the strokes.

The optical input element is preferably sensitive in such a way that it provides information about the lifting and flying of the pen in order to improve the recognition, in particular the pattern of the stroke, on the basis of this information.

In an alternative embodiment, the input element is capacitive in nature.

The method is preferably implemented by software. It is a

A computer program product. Part of the invention is also a data carrier that has the software.

The invention is explained in more detail below with the aid of exemplary embodiments which are illustrated in the figures. The same reference numbers in the individual figures denote the same elements. In detail show:

Figures 1-11 show a series of stroke inputs; Cd o> IV) 'M

Cπ o π o Cn O Cπ

3

3 rt

Φ

H cn

Ω f

F-

Φ

F ¹

F-

Ω f

Φ t

Φ

3

3 rr

N

Φ i

N

3

<

Φ

H

£

DJ

For

Φ

3

cυ Cd tv> tv> F F

Cπ o Cn o Cπ O Cπ s: 3 CΛ α t α pr ö N s: £ s. PL P Φ P Φ • u s: P Φ CΛ cn et φ tu <i ≥i CΛ σ φ CΛ 3 σ

3 P rt Φ F- F- li F- 3 F- F- Φ PH 3 li o PF ¹ F- rt o tr li F- o 3 F- φ F- * <; 3 F- li 3 HH 3 φ Φ φ <HH F- cn ω 3 CΛ 3 F ¹ HF "3 H Φ 3 3 rt 3 H 3 3 rt φ

PL O vQ F- o - rt Φ iQ F- vQ rt φ Φ os: ii F- Φ F ¹ N 3 tr Φ

Φ F- T tr P P w • U li Φ CΛ • U 0 Φ rt Φ Φ H cn r F- iQ rt CΛ Φ Φ α CΛ o li <!

• 3 Φ Φ tr pr ii F- P CJ. H << O rt iQ rt 1 rt φ »i P rt s: rt 3 3 ι rt o to H Φ rt 3 ι 3 3 φ Φ cn F- Φ Φ CΛ F- P- 3 Φ p: _^ Φ F- g Φ Φ ri α • U φ Hi 0 3 3 F- Hi rt 3 rt iQ P tr F ¹ rt PL 3 O: r ι F ¹ F- 3 F- ω tr F-

F- F- F- φ φ H rt N Φ Φ 3 Φ rt H F- vQ F "Φ cn r rt P s: 3 rt o Φ F ¹ F-

Φ iQ rt 3 F ¹ Φ Φ rt * U 3 3 iQ 3 Φ o φ to F "to φ iQ ^ Φ F- Φ Φ • i H rt Φ

3 Φ to ω F "3 Φ O 3 Φ pr F- Ω ii H ü iQ 3 cn O pr iQ

* H H rt φ • u Φ 3 P- CΛ ≤ 3 φ <! 3 iQ pr ω P O Φ Φ 3 pr t) φ ∑: Φ

F- <Q φ 3 F> 3 P iQ o rt P 3 o F- ii 3 Hi tr 3 φ 3 Φ 3 Q Cπ <i Φ H πd 3 α to Φ tr Φ 3 P 3 3 H 3 3 F- Ω PL «PL 3 1 IS! 3 H PL

3 o N ω O W tr <! H 1 H α H F- H Φ φ pr 3 3 = O N F- F- tu 3 rt Φ

H Φ 3 Φ to rt F- Φ F- t! CΛ Φ rt PL Ω Ω F- φ 3 tr rt tr 3 Φ Φ cn H.

Φ H F- P F- ω H rt O rt CΛ F-. P 3 "Pf on Φ Φ _^ Φ ^• d F- cn rt; r tr 3 <!

3 PT O Ω φ rt P CΛ rt H 3 3 rt Pf rt 3 3 o H o Ω Φ H Φ ω σ • o

Φ tr pf H F- 3 Ω Φ tr O α M 3 Φ rt Φ Φ H pr 3 e H O 3 to Φ H

CTι 3 Φ 3 O F> pr 3 F- pr P et cn F- F- 3 3 CΛ F- F- iQ P Φ Φ T 3 rt H σ H

3 3 Φ> U 3 tu F- 1 φ H tr rt 3 3 3 cn "Ta to 3 Φ 3 3 3 Φ Φ F- Φ F-

3 Φ rt F- 3 N Φ CΛ CΛ P Φ φ Φ KQ 3 rt M • d rt 3 to li w 3 Ω

3 3 3 Φ ιQ Φ rt rt pr 3 H P H 3 "U Φ tr • ω Φ rt Φ 3 F- 3 3 pr

P 3 3 Φ φ F- Φ H ü Φ iQ o P F- o ^ pr 3 to vQ rt Φ F- cl rt

Ω ii cn ιQ 3 OO 3 PPF ¹ 3 3 F ¹ s: ä φ Φ rt rt Φ Φ 3 ω 3 3

-j PL pr CΛ φ φ Φ Φ T T 3 3 3 φ 3 iQ Hi cn o tr 3 F- Φ Φ H to F- rt F- 3

P rt 3 CΛ F- 3 3 Φ Φ rt to vQ H vQ F- Φ 3 tr o 3 F- H O Ω W to iQ

N to 3 H rr 3 • Ω cn PO 1 3 O: to Φ F- 3 to f ^{^} rt PL O rt

Φ to 3 OF ¹ F- Φ φ ι ι N pr iQ F- li CΛ 3 Φ iQ rt F- cn ι Ω Φ HPT ü f ^

F- rt> r Hι ü F- O Φ Φ Sl F ¹ H 3 F- rt cn 3 F ¹ PP rt OS Pf 1 F- N w Φ OPQ a Φ Φ tr rr 3 s: F- Φ F- P iQ rt H rt F- ü 3 • F> 3 Φ tu Φ 3 1 T 3

Φ Φ 3 cn F- Φ tr ⁴ Φ iQ P Ω 3 Φ Hi cn M o Φ 3 Ω rt Φ to iQ F- H tr 3 3 tu Φ 3

3 F,. CΛ cn F- 3 Φ ü Pf 3 α Φ 3? H p: tr Φ li N S! Pr rt pr Φ p: 3 ü 1

<• 3 3 rr Φ Φ 3 PL 3 Φ Ω Φ rt 3 Φ 1 F- Φ Φ 3 Ω H pr t F- t Φ ≥! F ¹ F- tr 3 Φ 3 P. 3 Φ cn pr ü. u * ⁱ Q 3 CΛ ω 3 3 pr Φ H 3

F- 3 HPF ¹ H Φ N cn rr F- α PH tr O cn F "Φ 3 rt rt 3 £ cn P 3 pr

Φ Ω Ω 3 S rt P Φ 3 P Φ H rt ¹ U ti iQ tr HH t Φ 3 rt 3 3 Φ Hi

Pf F- Pf cn tr «Φ 3 3 cn 3 Hi * U 0 CΛ pr φ O F- O Φ P. OOPHH to Φ 3 O tu ω Φ F- o NH 3 cn o iQ O Hl Ω P 3 F ¹ Φ H 3 F- pr PT 3 PL 3 H 3 F "

F- rt Hl α Φ tr s: 3 <! iQ 3 Ω Φ F- tr 3 vQ iQ F- Φ φ Φ 3 3 φ F- φ Φ vQ

3 P **. Φ Pf Φ φ P Φ F- 3 ^J Hi iQ H CΛ Φ rt rt CΛ 1 3. rt F- F- H Φ Q tr H rt li Pf H iQ Ω 0- Φ Φ F- φ rt 1 ϊ_j ^J P, Hi t < _^ «3 cn" _^ 3

P Φ to Φ Φ H F- P Φ - 3 pf 1 3 Φ 3 H CΛ F- 3 P 3 0- HO σ φ rt PL tr F- t 3 3 F- Φ P 3 tr rt F- H CΛ tr _^ Location 3 3 to F- H pr P £ 3 Φ

Φ t φ F- P 3 H 3 3 Φ 3 rt rt P Φ pr ü ω Ω φ F "tr Φ P Φ f 3 3 tu Φ Φ rt 3 P- H pr 3 s: φ OO PL pr 3 to Φ CΛ 3 Ü «φ Φ iQ Φ ω Φ p:" »Φ Φ O rt φ F- • pr tr tr 0 rt Φ 3 rt F- Ω rt Ω O

Φ F- P PL H CΛ 3 Pf Ü cn to pr 0 H φ φ Φ F, 3 Φ p: pr HT 3 3 cn 3 "U tr F- Φ ω rt 3 £ rt O Φ Φ CΛ P FI ü F- Ω PL H 3 F- φ jo 3 • d

Ω _ ,. Φ Φ F- rt H <F "Φ ιu φ F ¹ * <J cn <iQ pr Φ Φ to cn pr F- <P pr Φ iQ Ω ti O Φ F- 3 P 3 3: rt F" 3 PL rt o PHH -J F- rt φ 3 o 3 iQ 3 Φ s pr OTH Ω 3 F "cn tr P Φ tr PH 3 Φ Φ 3 iQ CΛ Φ

Φ Φ H F- PT Φ Hi Pf φ ci ^¬ 3 o cn PL - iQ ü F- • U F- Φ Φ ^ tr -> 3

F- tr 3 tr Φ CΛ P Φ H to F <F- to 3 P rt to Φ Φ cn rt

3 Φ J-. Φ F- Φ CΛ • pr £ o F- F- * d p: CΛ N Φ cn iQ t Φ rt <! tr m H rt Φ

Φ 3 _^ »CΛ F ¹ cn l F- Ω cn 3 H Ω rt 3 cn o H F- 3 F- φ O <lo rt D: 3

3 N σ φ Φ pr rt Φ 0: pr H <! Φ P 3 3 OF "φ tr ¹ 3

? 3 φ Φ 3 3 Hi O O ii Hi iQ F- rt 3 pr P

P 3 F- ti • cn rt F-? H Φ P CΛ Φ F- rt iQ F-

3 iQ ιQ O to Ω Φ 3 tr rt 3 3 • Φ iQ

3 rt o tr Φ " ^» * _* Φ

Letter a. Figures 8 to 11 show the input of the letters F and a.

Figures 12-21 show the entry of a stroke row. Before the stroke row is output in the form of a word, each stroke is shown in a display element 4 and thus reported back to the user. Until the end of a word, the user therefore only sees his own hand-drawn strokes, which he has entered on an input element 5. However, the invention is able to tell the user the most likely hypothesis based on these strokes. That the graphic representation of the strokes shows which word it will be. The corresponding word is then output at the end of the entry. In an embodiment that is not shown, the user is offered a selection of symbols or words to be displayed. He can then use the stylus to select the corresponding symbol without having to complete the entry. Figures 21 and 22 show that a complete word can also serve as a symbol. A symbol does not always have to stand for a letter.

An increase in performance can be achieved through special input elements. They must be able to approach the donor or remove the donor

Provide pen information. In a further step, the direction in the short phase of the still visible lifting of the pen tip and the correspondingly short entry phase before the next stroke could be included in the detection.

In a further embodiment e.g. the recognition process is supported by an acceleration recognition.

FIG. 22 shows the flow diagram of the recognition process. At the beginning the status is on start set. A stroke is read. Next, it is checked whether the pattern created by the stroke is a point, a line, or some other stroke. A point and a line can be interpreted as control characters. Control characters are used to determine whether a letter should be capitalized or whether a certain function should be carried out. Depending on the input, the status is set to either Status + Stroke, Status + Line or Status + Point.

The next step is to check whether the status has the value Start. If this is the case, it is checked whether the pattern of the read stroke is listed in a transition graph as the start element. This means that a transition graph is sought that has this pattern as the start element. If a transition graph is found that has this start element, it is checked whether there are other elements in this graph. If, on the other hand, no transition graph is found that has this start element, it is checked whether a symbol can be found that is described by the stroke or its pattern.

If, on the other hand, the status was not set to start, it is checked whether there is another element in the selected transition graph that matches the pattern of the stroke. If there is no element in the transition graph that matches the entered pattern, it is checked whether there is a transition graph that contains the previously read strokes or their patterns u. Of course, this can only be the case if several transition graphs overlap with respect to their first elements. The previous transition graphs must therefore be checked to see whether they have been completely covered by the previous patterns. If this is the case, the patterns that are not considered Cπ O Cπ pr θ:

3

Φ

3

Claims

claims

1. Method for handwriting recognition, wherein a symbol is determined by patterns of strokes and one or more transition graphs,

- with a first step, in which a first stroke is analyzed to determine its pattern,

with a second step, in which a transition graph is determined which has the pattern determined in the first step,

with a third step, in which further strokes are analyzed as long as the patterns of these strokes are contained in the examined transition graph,

- With a fourth step, in which the symbol is determined, that determined in the third step

Transitional graph is assigned.

2. The method according to claim 1, characterized in that the transition graph is preferably completely described by the entered strokes.

3. The method according to claim 1 or 2, characterized in that in the event of detection of a stroke, the pattern of which is not contained in the examined transition graph, there is a return to the previously recognized pattern until a transition graph is found which is characterized by the up to this pattern is completely covered in order to then determine the corresponding symbol.

4. The method according to claim 3, characterized in that the strokes, to which no transition graph could be assigned on the return, are used for determining a new transition graph.

5. Method for handwriting recognition of symbols, a symbol being determined by a pattern of a stroke or by a plurality of patterns in connection with a transition graph, with a first step in which a stroke is read and assigned to a pattern,

with a second step, in which it is checked whether the pattern is a new entry or a continuation of a previous entry,

with a third step, in which, in the event that it is not a start pattern in a transition graph, an attempt is made to determine a symbol for the new input, and in the event that it is the Continuation of an input, an attempt is made to determine a symbol based on the previously recognized transition graph,

- with a fourth step, in the event that there is a further pattern in the transition graph or a start pattern in the transition graph, and the

The transition graph has no further patterns, the symbol is determined, and in the event that there is a further pattern in the transition graph or a starting pattern in the transition graph and the transition graph has further patterns, proceed to step one.

6. The method according to claim 5, characterized in that a status can be used to determine whether it is the new input or the continuation of the input, the status being set to start whenever a symbol has been determined or a An attempt has been made to determine the symbol.

7. The method according to one or more of claims 1 to 6, characterized in that the symbol is output on a screen.

8. The method according to one or more of claims 1 to 7, characterized in that the recognized symbols are only output on the display when a word has been completely ended, the previously entered Strokes, graphically represented so that the recognized word can be derived from them.

9. The method according to claim 8, characterized in that word suggestions are provided before completion of the input.

10. The method according to one or more of claims 1 to

9, characterized in that the transition graph and the patterns can be dynamically assigned to each symbol in a configuration step, preferably depending on a user.

11. The method according to one or more of claims 1 to

10, characterized in that the assignment of the strokes to a transition graph depends on the time interval of the input of the strokes.

12. The method according to one or more of claims 1 to

11, characterized in that the transition graph determines the order of the strokes to be entered.

13. The method according to one or more of claims 1 to

12, characterized in that a symbol for a letter and / or an entire word and / or more

Words and / or other graphic elements can stand.

14. Device, in particular a mobile input device, for handwriting recognition of symbols, a symbol being determined by patterns of strokes and / or a plurality of transition graphs, with a means for recognizing strokes and with a processing unit, characterized in that the processing unit includes the strokes with assigns a method according to one or more of claims 1 to 11 symbols in order to then supply the symbols for further processing.

15. The apparatus according to claim 14, characterized in that the means for detecting strokes is an optical ^' Is an input element for pens or a pen-like input device.

16. The apparatus according to claim 15, characterized in that the optical input element is so sensitive that it provides information about the lifting and / or flying and / or accelerating the pen in order to use this information to identify, in particular the pattern of the stroke improve.

17. The apparatus according to claim 14, characterized in that the optical input element is capacitive.

18. The device according to one or more of claims 14 to 17, characterized in that the method is mapped by software on the hardware.