US4718102A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Process and apparatus involving pattern recognition
Assignee: Communication Intelligence Corp
•
Filing: 1987-04-03
•
Granted: 1988-01-05
•
Expires: 2007-04-03
Abstract
Abstract Processing of handwritten patterns, for example, inputting of complex characters, such as Kanji, into a word processor involves a balance between an algorithmic part and a disambiguation part. Application of routines uniformly during the algorithmic part deliberately results in confusion sets of characters which are subsequently differentiated during the disambiguation part. Balance between the two parts permits a flexibility to accommodate differences in writing styles, while application of routines customized to the established character patterns permits the unique solution necessary for accurate recognition of a character.
Inventors
- 👤 Hewitt D. Crane
- 👤 John S. Ostrem
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-26 11:18
CPC Classifications
G; G06; G06V; G06V30/00; G06V30/10; G06V30/24; G06V30/242; G06V30/32; G06V30/333; G06V30/36; G06V30/373
Claims Analysis
Total Claims: 64
Claim 1
Independent Claim
263 words
Apparatus for pattern processing, said apparatus comprising means for receiving first signal information derived from information comprising handwritten patterns, and data processing means provided with prestored second signal information for uniquely differentiating said handwritten patterns from patterns within a pattern vocabulary, said vocabulary being made up of one or more vocabulary sets with a set defining patterns with at least one common characteristic within which differentiation is restricted during at least a portion of processing, said second signal information being arranged to accomplish pattern processing by routines comprising algorithmic routines in which all patterns within at least a vocabulary set are processed similarly and disambiguation routines in which patterns are processed in terms of characteristics of specific patterns, in which use of algorithmic routines alone without disambiguation routines would yield confusion sets, said sets consisting of patterns confused with each other and in which use of both algorithmic routines and disambiguation routines yields uniquely differentiated patterns and second signal information including a dictionary containing the vocabulary and provision for matching information derived from said first signal information with dictionary entries within the appropriate set, characterized in that said algorithmic routines if used alone without any disambiguation routines results in at least 50 percent of all patterns within the pattern vocabulary being in confusion sets, each said set containing an averave of at least two patterns, and in which the said disambiguation routines include non-uniform, customized routines selected to distinguish members within confusion sets, while at the same time surviving handwriting variations, whereby said data processing means comprises means for producing unique character recognition.
Claim 2
Dependent Claim
263 words
Apparatus for pattern processing, said apparatus comprising means for receiving first signal information derived from information comprising handwritten patterns, and data processing means provided with prestored second signal information for uniquely differentiating said handwritten patterns from patterns within a pattern vocabulary, said vocabulary being made up of one or more vocabulary sets with a set defining patterns with at least one common characteristic within which differentiation is restricted during at least a portion of processing, said second signal information being arranged to accomplish pattern processing by routines comprising algorithmic routines in which all patterns within at least a vocabulary set are processed similarly and disambiguation routines in which patterns are processed in terms of characteristics of specific patterns, in which use of algorithmic routines alone without disambiguation routines would yield confusion sets, said sets consisting of patterns confused with each other and in which use of both algorithmic routines and disambiguation routines yields uniquely differentiated patterns and second signal information including a dictionary containing the vocabulary and provision for matching information derived from said first signal information with dictionary entries within the appropriate set, characterized in that said algorithmic routines if used alone without any disambiguation routines results in at least 50 percent of all patterns within the pattern vocabulary being in confusion sets, each said set containing an averave of at least two patterns, and in which the said disambiguation routines include non-uniform, customized routines selected to distinguish members within confusion sets, while at the same time surviving handwriting variations, whereby said data processing means comprises means for producing unique character recognition.
Claim 3
Dependent Claim
57 words
Apparatus of claim 1 in which data processing means is designed to result in "forward processing" in which algorithmic routines are applied initially in an algorithmic part which terminates in dictionary matching to actually yield said confusion sets followed by disambiguation routines in a non-algorithmic part to yield a uniquely differentiated pattern within each said confusion set.
Claim 4
Dependent Claim
23 words
Apparatus of claim 1 in which processing means in designed such that at least one disambiguation routine precedes at least one algorithmic routine.
Claim 5
Dependent Claim
12 words
Apparatus of claim 3 in which confusion sets are not actually yielded.
Claim 6
Dependent Claim
16 words
Apparatus of claim 1 in which confusion sets contain an average of at least three patterns.
Claim 7
Dependent Claim
16 words
Apparatus of claim 1 in which confusion sets contain an average of at least four patterns.
Claim 8
Dependent Claim
44 words
Apparatus of claim 1 in which the said pattern vocabulary contains at least 2,000 patterns and in which at least 90 percent of all patterns in the pattern vocabulary are in confusion sets, said confusion sets containing an average of at least 5 patterns.
Claim 9
Dependent Claim
45 words
Apparatus of claim 7 in wich the pattern vocabulary contains at least 3,000 patterns and in which at least 14 out of 15 of all patterns in the pattern vocabulary are in confusion groups, said confusion sets containing an average of at least 71/2 patterns.
Claim 10
Dependent Claim
36 words
Apparatus of claim 1 in which the pattern vocabulary contains in excess of 800 patterns and in which the minimum number of patterns per confusion set is increased linearly as the vocabulary size increases above 800.
Claim 11
Dependent Claim
31 words
Apparatus of any one of claims 1 through 9 in which the minimum number of patterns per confusion set is increased by 50 percent over the minima expressed in such claims.
Claim 12
Dependent Claim
31 words
Apparatus of any one of claims 1 through 9 in which the minimum number of patterns per confusion set is increased by 100 percent over the minima expressed in such claims.
Claim 13
Dependent Claim
18 words
Apparatus to claim 1 in which said means for receiving first signal information comprises a pattern registration means.
Claim 14
Dependent Claim
19 words
Apparatus of claim 12 in which said first signal information is coordinate information consisting of pluralities of spaced points.
Claim 15
Dependent Claim
32 words
Apparatus of claim 13 in which points are equally temporally spaced and in which the said data processing means converts such first signal information into points separated by equally spaced distance intervals.
Claim 16
Dependent Claim
13 words
Apparatus of claim 1 in which said first signal information includes stroke information.
Claim 17
Dependent Claim
13 words
Apparatus of claim 15 in which said second signal information includes stroke information.
Claim 18
Dependent Claim
16 words
Apparatus of claim 16 in which such stroke information includes permitted tolerances on a point-by-point basis.
Claim 19
Dependent Claim
13 words
Apparatus of claim 17 in which such stroke information is in angular terms.
Claim 20
Dependent Claim
26 words
Apparatus of claim 18 in which said data processing means includes preprocessing means for recognizing strokes which commonly occur in patterns within the said pattern vocabulary.
Claim 21
Dependent Claim
28 words
Apparatus of claim 19 in which said preprocessing means recognizes strokes which approximate straight lines over some substantial portion of their length, such strokes being designated "simple strokes".
Claim 22
Dependent Claim
31 words
Apparatus of claim 20 in which the data processing means provides for labeling of any strokes not recognized by the preprocessing means by means of templates containing point-by-point permitted tolerance values.
Claim 23
Dependent Claim
13 words
Apparatus of claim 1 in which the algorithmic routines include stroke string processes.
Claim 24
Dependent Claim
25 words
Apparatus of claim 22 in which said disambiguation routines includes correlational processes for differentiating characters within confusion sets from all characters within a vocabulary set.
Claim 25
Dependent Claim
21 words
Apparatus of claim 1 in which said data processing means includes means for further processing information subsequent to unique pattern recognition.
Claim 26
Dependent Claim
21 words
Apparatus of claim 24 in which said means for further processing includes formalized replication of an input character within a medium.
Claim 27
Dependent Claim
10 words
Apparatus of claim 25 designed to function as a typesetter.
Claim 28
Dependent Claim
10 words
Apparatus of claim 25 designed to function as a typewriter.
Claim 29
Dependent Claim
13 words
Apparatus of claim 24 in which said further processing means comprises a computer.
Claim 30
Dependent Claim
13 words
Apparatus of claim 28 in which said computer functions as a word processor.
Claim 31
Dependent Claim
23 words
Apparatus of claim 1 in which the said dictionary contains pattern variations so that the number of dictionary entries exceeds said pattern vocabulary.
Claim 32
Dependent Claim
49 words
Apparatus of claim 30 in which the said first signal information includes strokes, a stroke being defined as a pattern element resulting from continuous contact of pattern forming means and pattern accommodating means, as well as stroke sequence information and in which said pattern variations include stroke sequence variations.
Claim 33
Dependent Claim
259 words
Method for pattern processing comprising comparison of input patterns with prestored second signal information in which first signal information derived from information comprising handwritten patterns is input and compared with prestored second information to uniquely differentiate said handwritten patterns from other patterns within a pattern vocabulary, said vocabulary being made up of one or more vocabulary sets, each said set defining patterns with at least one common characteristic within which differentiation is restricted during at least a portion of processing, said pattern processing being by routines comprising algorithmic routines in which all patterns within at least a vocabulary set are processed similarly and disambiguation routines in which patterns are processed in terms of characteristics selected specifically to distinguish numbers of confusion sets while at the same time surviving handwriting variations in which use of algorithmic routines alone without disambiguation routines would yield confusion sets together containing at least 50% of all patterns within the pattern vocabulary, said sets consisting of patterns confused with each other and in which use of both algorithmic routines and disambiguation routines yields uniquely differentiated patterns, said second signal information including a dictionary containing the vocabulary and matching information derived from said first signal information with dictionary entries within the appropriate set, characterized in that said algorithmic routines, if used alone without any disambiguation routines results in at least 50 percent of all patterns within the pattern vocabulary being in confusion sets with each said set containing an average of at least two patterns, whereby said data processing comprises means for producing unique character recognition.
Claim 34
Dependent Claim
259 words
Method for pattern processing comprising comparison of input patterns with prestored second signal information in which first signal information derived from information comprising handwritten patterns is input and compared with prestored second information to uniquely differentiate said handwritten patterns from other patterns within a pattern vocabulary, said vocabulary being made up of one or more vocabulary sets, each said set defining patterns with at least one common characteristic within which differentiation is restricted during at least a portion of processing, said pattern processing being by routines comprising algorithmic routines in which all patterns within at least a vocabulary set are processed similarly and disambiguation routines in which patterns are processed in terms of characteristics selected specifically to distinguish numbers of confusion sets while at the same time surviving handwriting variations in which use of algorithmic routines alone without disambiguation routines would yield confusion sets together containing at least 50% of all patterns within the pattern vocabulary, said sets consisting of patterns confused with each other and in which use of both algorithmic routines and disambiguation routines yields uniquely differentiated patterns, said second signal information including a dictionary containing the vocabulary and matching information derived from said first signal information with dictionary entries within the appropriate set, characterized in that said algorithmic routines, if used alone without any disambiguation routines results in at least 50 percent of all patterns within the pattern vocabulary being in confusion sets with each said set containing an average of at least two patterns, whereby said data processing comprises means for producing unique character recognition.
Claim 35
Dependent Claim
49 words
Method of claim 32 in which "forward processing" is applied which results in algorithmic routines which are applied initially in an algorithmic part which terminates in dictionary matching to actually yield said confusion sets followed by disambiguation routines to yield a uniquely differentiated pattern within each said confusion set.
Claim 36
Dependent Claim
17 words
Method of claim 32 in which at least one disambiguation routine precedes at least one algorithmic routine.
Claim 37
Dependent Claim
12 words
Method of claim 34 in which confusion sets are not actually yielded.
Claim 38
Dependent Claim
16 words
Method of claim 32 in which confusion sets contain an average of at least three patterns.
Claim 39
Dependent Claim
16 words
Method of claim 32 in which confusion sets contain an average of at least four patterns.
Claim 40
Dependent Claim
43 words
Method of claim 32 in which said pattern vocabulary contains at least 2,000 patterns and in which at least 90 percent of all patterns in the pattern vocabulary are in confusion sets, said confusion sets containing an average of at least 5 patterns.
Claim 41
Dependent Claim
46 words
Method of claim 38 in which the pattern vocabulary contains at least 3,000 patterns and in which at least 14 out of 15 of all patterns in the pattern vocabulary are in confusion groups with such confusion groups containing an average of at least 71/2 patterns.
Claim 42
Dependent Claim
36 words
Method of claim 32 in which the pattern vocabulary contains in excess of 800 patterns and in which the minimum number of patterns per confusion set is increased linearly as the vocabulary size increases above 800.
Claim 43
Dependent Claim
31 words
Method of any one of claims 32 through 40 in which the minimum number of patterns per confusion set is increased by 50 percent over the minima expressed in such claims.
Claim 44
Dependent Claim
31 words
Method of any one of claims 32 through 40 in which the minimum number of patterns per confusion set is increased by 100 percent over the minima expressed in such claims.
Claim 45
Dependent Claim
15 words
Method of claim 32 in which first signal information is derived from pattern registration means.
Claim 46
Dependent Claim
19 words
Method of claim 43 in which said first signal information is coordinate information consisting of pluralities of spaced points.
Claim 47
Dependent Claim
21 words
Method of claim 44 in which said first signal information is converted to yield points separated by equally spaced distance intervals.
Claim 48
Dependent Claim
12 words
Method of claim 32 in which said first signal information defines strokes.
Claim 49
Dependent Claim
13 words
Method of claim 46 in which said second signal information includes stroke information.
Claim 50
Dependent Claim
16 words
Method of claim 46 in which such stroke information includes permitted tolerances on a point-by-point basis.
Claim 51
Dependent Claim
13 words
Method of claim 48 in which such stroke information is in angular terms.
Claim 52
Dependent Claim
25 words
Method of claim 49 in which said first signal information is preprocessed to recognizing strokes which commonly occur in patterns within the said pattern vocabulary.
Claim 53
Dependent Claim
29 words
Method of claim 50 in which said preprocessing means recognizes "simple strokes" which approximate straight lines over some substantial portion of their length, such strokes being designated "simple strokes".
Claim 54
Dependent Claim
22 words
Method of claim 51 in which strokes not recognized during preprocessing are labeled by comparison with templates containing point-by-point permitted tolerance values.
Claim 55
Dependent Claim
13 words
Method of claim 32 in which said algorithmic routines include stroke string processes.
Claim 56
Dependent Claim
25 words
Method of claim 53 in which said disambiguation routines include correlational processes for differentiating characters within confusion sets from all characters within a vocabulary set.
Claim 57
Dependent Claim
13 words
Method of claim 32 in which processing continues subsequent to unique pattern recognition.
Claim 58
Dependent Claim
21 words
Method of claim 55 in which further processing includes replicating an input pattern corresponding with first signal information within a medium.
Claim 59
Dependent Claim
6 words
Method of claim 55 comprising typesetting.
Claim 60
Dependent Claim
6 words
Method of claim 56 comprising typewriting.
Claim 61
Dependent Claim
12 words
Method of claim 55 in which further processing is in a computer.
Claim 62
Dependent Claim
11 words
Method of claim 55 in which further processing comprises word processing.
Claim 63
Dependent Claim
23 words
Method of claim 32 in which the said dictionary contains pattern variations so that the number of dictionary entries exceeds said pattern vocabulary.
Claim 64
Dependent Claim
49 words
Method of claim 61 in which the said first signal information includes strokes, a stroke being defined as a pattern element resulting from continuous contact of pattern forming means and pattern accommodating means, as well as stroke sequence information and in which said pattern variations include stroke sequence variations.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1991-01-04 | 1991-07-06 | 1992-01-05 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1995-01-04 | 1995-07-06 | 1996-01-05 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 1999-01-04 | 1999-07-06 | 2000-01-05 | $7700.00 | Window Closed / Lapsed |