Patent US2943757A successfully refreshed from Google Patents.
Patent US5343537A successfully refreshed from Google Patents.
US5343537A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Statistical mixture approach to automatic handwriting recognition
Assignee: International Business Machines Corp
•
Filing: 1991-10-31
•
Granted: 1994-08-30
•
Expires: 2011-10-31
Abstract
Abstract Method and apparatus for automatic recognition of handwritten text based on a suitable representation of handwriting in one or several feature vector spaces(s), Gaussian modeling in each space, and mixture decoding to take into account the contribution of all relevant prototypes in all spaces. The feature vector space(s) is selected to encompass both a local and a global description of each appropriate point on a pen trajectory. Windowing is performed to capture broad trends in the handwriting, after which a linear transformation is applied to suitably eliminate redundancy. The resulting feature vector space(s) is called chirographic space(s). Gaussian modeling is performed to isolate adequate chirographic prototype distributions in each space, and the mixture coefficients weighting these distributions are trained using a maximum likelihood framework. Decoding can be performed simply and effectively by accumulating the contribution of all relevant prototype distributions. Post-processing using a language model may be included.
Inventors
- 👤 Eveline J. Bellegarda
- 👤 Jerome R. Bellegarda
- 👤 David Nahamoo
- 👤 Krishna S. Nathan
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-29 19:42
CPC Classifications
G; G06; G06V; G06V30/00; G06V30/10; G06V30/32; G06V30/333; G06V30/347; G06F; G06F18/00; G06F18/20; G06F18/21; G06V30/36; G06V30/373
Claims Analysis
Total Claims: 14
Claim 1
Independent Claim
275 words
A handwriting recognition system comprising: an electronic tablet for measuring the value of at least one feature of a handwriting stroke on the tablet at each of a plurality of different positions along the stroke to produce a set of feature vector signals representing the handwriting stroke; means for storing a plurality of prototype signals, each prototype signal having at least one parameter value; means for storing, for each character in an alphabet of characters and for each prototype signal, an estimate of the probability of occurrence of the prototype signal given the occurrence of the character; means for generating a component character match score for a character and for a feature vector signal and a prototype signal, said component character match score comprising an estimate of the probability of occurrence of the feature vector signal given the occurrence of the prototype signal, combined with the probability of occurrence of the prototype signal given the occurrence of the character; means for generating a partial character match score for a character and a feature vector signal, said partial character match score comprising a combination of the component character match scores for the character and the feature vector signal for all prototype signals; means for generating a full character match score for a character and for all feature vector signals representing the handwriting stroke, said full character match score comprising a combination of the partial character match scores for the character and for all feature vector signals in the handwriting stroke; means for identifying at least one best candidate character having the best full character match score; and means for outputting at least one best candidate character.
Claim 2
Dependent Claim
275 words
A handwriting recognition system comprising: an electronic tablet for measuring the value of at least one feature of a handwriting stroke on the tablet at each of a plurality of different positions along the stroke to produce a set of feature vector signals representing the handwriting stroke; means for storing a plurality of prototype signals, each prototype signal having at least one parameter value; means for storing, for each character in an alphabet of characters and for each prototype signal, an estimate of the probability of occurrence of the prototype signal given the occurrence of the character; means for generating a component character match score for a character and for a feature vector signal and a prototype signal, said component character match score comprising an estimate of the probability of occurrence of the feature vector signal given the occurrence of the prototype signal, combined with the probability of occurrence of the prototype signal given the occurrence of the character; means for generating a partial character match score for a character and a feature vector signal, said partial character match score comprising a combination of the component character match scores for the character and the feature vector signal for all prototype signals; means for generating a full character match score for a character and for all feature vector signals representing the handwriting stroke, said full character match score comprising a combination of the partial character match scores for the character and for all feature vector signals in the handwriting stroke; means for identifying at least one best candidate character having the best full character match score; and means for outputting at least one best candidate character.
Claim 3
Dependent Claim
18 words
A handwriting recognition system as claimed in claim 1, characterized in that the output means comprises a display.
Claim 4
Dependent Claim
49 words
A handwriting recognition system as claimed in claim 2, characterized in that the full character match score for a candidate character further comprises an estimate of the probability of occurrence of the candidate character given the occurrence of one or more other characters preceding or following the candidate character.
Claim 5
Dependent Claim
66 words
A handwriting recognition system as claimed in claim 3, characterized in that: each prototype signal comprises at least one mean parameter value and at least one variance parameter value; and the probability of occurrence of a feature vector signal given the occurrence of a prototype signal is estimated as the Gaussian likelihood of the feature vector signal given the mean and variance of the prototype signal.
Claim 6
Dependent Claim
48 words
A handwriting recognition system as claimed in claim 3, characterized in that the electronic tablet measures the value of at least one local feature of a handwriting stroke on the tablet, and measures the value of at least one global feature of a handwriting stroke on the tablet.
Claim 7
Dependent Claim
110 words
A handwriting recognition system as claimed in claim 5, characterized in that: the component character match score comprises an estimate of the product of the probability of occurrence of the feature vector signal given the occurrence of the prototype signal, multiplied by the probability of occurrence of the prototype signal given the occurrence of the character; the partial character match score comprises the sum of the component character match scores for the character and the feature vector signal for all prototype signals; and the full character match score comprises the product of the partial character match scores for the character and for all feature vector signals in the handwriting stroke.
Claim 8
Dependent Claim
261 words
A handwriting recognition method comprising the steps of: measuring the value of at least one feature of a handwriting stroke on an electronic tablet at each of a plurality of different positions along the stroke to produce a set of feature vector signals representing the handwriting stroke; storing a plurality of prototype signals, each prototype signal having at least one parameter value; storing, for each character in an alphabet of characters and for each prototype signal, an estimate of the probability of occurrence of the prototype signal given the occurrence of the character; generating a component character match score for a character and for a feature vector signal and a prototype signal, said component character match score comprising an estimate of the probability of occurrence of the feature vector signal given the occurrence of the prototype signal, combined with the probability of occurrence of the prototype signal given the occurrence of the character; generating a partial character match score for a character and a feature vector signal, said partial character match score comprising a combination of the component character match scores foe the character and the feature vector signal for all prototype signals; generating a full character match score for a character and for all feature vector signals representing the handwriting stroke, said full character match score comprising a combination of the partial character match scores for the character and for all feature vector signals in the handwriting stroke; identifying at least one best candidate character having the best full character match score; and outputting at least one best candidate character.
Claim 9
Dependent Claim
261 words
A handwriting recognition method comprising the steps of: measuring the value of at least one feature of a handwriting stroke on an electronic tablet at each of a plurality of different positions along the stroke to produce a set of feature vector signals representing the handwriting stroke; storing a plurality of prototype signals, each prototype signal having at least one parameter value; storing, for each character in an alphabet of characters and for each prototype signal, an estimate of the probability of occurrence of the prototype signal given the occurrence of the character; generating a component character match score for a character and for a feature vector signal and a prototype signal, said component character match score comprising an estimate of the probability of occurrence of the feature vector signal given the occurrence of the prototype signal, combined with the probability of occurrence of the prototype signal given the occurrence of the character; generating a partial character match score for a character and a feature vector signal, said partial character match score comprising a combination of the component character match scores foe the character and the feature vector signal for all prototype signals; generating a full character match score for a character and for all feature vector signals representing the handwriting stroke, said full character match score comprising a combination of the partial character match scores for the character and for all feature vector signals in the handwriting stroke; identifying at least one best candidate character having the best full character match score; and outputting at least one best candidate character.
Claim 10
Dependent Claim
30 words
A handwriting recognition method as claimed in claim 7, characterized in that the step of outputting at least one best candidate character comprises displaying at least one best candidate character.
Claim 11
Dependent Claim
49 words
A handwriting recognition method as claimed in claim 8, characterized in that the full character match score for a candidate character further comprises an estimate of the probability of occurrence of the candidate character given the occurrence of one or more other characters preceding or following the candidate character.
Claim 12
Dependent Claim
66 words
A handwriting recognition method as claimed in claim 9, characterized in that: each prototype signal comprises at least one mean parameter value and at least one variance parameter value; and the probability of occurrence of a feature vector signal given the occurrence of a prototype signal is estimated as the Gaussian likelihood of the feature vector signal given the mean and variance of the prototype signal.
Claim 13
Dependent Claim
50 words
A handwriting recognition method as claimed in claim 10, characterized in that the step of measuring comprises measuring the value of at least one local feature of a handwriting stroke on the tablet, and measuring the value of at least one global feature of a handwriting stroke on the tablet.
Claim 14
Dependent Claim
110 words
A handwriting recognition method as claimed in claim 12, characterized in that: the component character match score comprises an estimate of the product of the probability of occurrence of the feature vector signal given the occurrence of the prototype signal, multiplied by the probability of occurrence of the prototype signal given the occurrence of the character; the partial character match score comprises the sum of the component character match scores for the character and the feature vector signal for all prototype signals; and the full character match score comprises the product of the partial character match scores for the character and for all feature vector signals in the handwriting stroke.
References Cited (Backward Citations)
JPS6015791A
Cited
JPS6029767A
Cited
US4731857A
Cited
JPS6474345A
Cited
US5138668A
Cited
JPH01321591A
Cited
US5075896A
Cited
US5029223A
Cited
US5105468A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1997-08-29 | 1998-02-28 | 1998-08-30 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2001-08-29 | 2002-02-28 | 2002-08-30 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2005-08-29 | 2006-02-28 | 2006-08-30 | $7700.00 | Window Closed / Lapsed |