Image analysis – Pattern recognition – Unconstrained handwriting
Reexamination Certificate
1996-06-21
2001-04-17
Chang, Jon (Department: 2723)
Image analysis
Pattern recognition
Unconstrained handwriting
C382S229000
Reexamination Certificate
active
06219449
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a character recognition system. More specifically, the present invention relates to a character recognition system receiving hand written characters as inputs, recognizing the character data thereof and outputting the result of recognition.
2. Description of the Background Art
FIG. 10
is a schematic block diagram of a conventional character recognition system. Referring to
FIG. 10
, input means
1
receives a hand written character as an input and turns it into electronic data. The input character data is applied to recognizing means
2
. Recognizing means
2
has necessary information for recognition, and recognizes the character data supplied from input means
1
. Results and states of input means
1
and recognizing means
2
are applied to output means
3
which display necessary information.
When hand written characters are to be recognized by the conventional character recognition system shown in
FIG. 10
, it is difficult to obtain high ratio of recognition, since hand written characters generally have much variations and omissions. Accordingly, in the conventional character recognition system, the method of recognition itself has been improved by developing a dictionary for recognition of characters which are likely to have recognition errors. In order to attain higher ratio of recognition, it is necessary to correct the result of recognition by using means other than those providing such result of recognition.
One representative method is correction of the obtained result of recognition by confirming grammatical meaning by using grammatical knowledge. In this method, correct word sets are identified by using grammatical knowledge based on a dictionary related to grammar. More specifically, a plurality of candidates for characters obtained as a result of recognition are combined appropriately and a combination of characters having proper grammatical meaning is found as a unit of an idiom or a phrase.
However, a large number of character combinations must be studied by using enormous knowledge related to grammar in order to obtain a combination having proper grammatical meaning. Meanwhile, the obtained candidates for characters are treated as equally weighted. Especially when hand written characters are recognized, the result of recognition is not very reliable. Therefore, even if first candidates for characters are given as a result of recognition, actually the reliability of recognition may possibly vary widely character by character. Accordingly, if the plurality of candidates for characters obtained as a result of recognition are simply combined one after another, a large number of combinations must be generated, taking a long period of time, in order to obtain a combination having a proper grammatical meaning.
SUMMARY OF THE INVENTION
Therefore, an object of the present invention is to provide a character recognition system capable of obtaining a combination of candidates for characters efficiently by generating a smaller number of combinations.
Briefly stated, in the present invention, input means receives hand written characters as inputs and turn them into electronic data; when thus generated data of characters are provided, variations of the input characters are expected based on the position of the data of the characters; and character recognition is carried out with the recognition results being ranked with priority based on the expected variations of the characters.
Therefore, according to the present invention, combinations of words are generated based on the reliability of the recognition results taking into consideration the expected reliability of variation, and grammatical meaning is confirmed. Therefore, the number of words to be examined and the necessary time for examination can be reduced.
In a more preferred embodiment of the present invention, information for determining priority of the recognition results is stored, and priority of each character is determined based on the stored information, dependent on the position of the input data of characters.
The foregoing and other objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.
REFERENCES:
patent: 4731857 (1988-03-01), Tappert et al.
patent: 4979227 (1990-12-01), Mittelbach et al.
patent: 4989258 (1991-01-01), Takahashi et al.
patent: 5018217 (1991-05-01), Yoshida et al.
patent: 5075896 (1991-12-01), Wilcox et al.
patent: 5151950 (1992-09-01), Hullender
patent: 5159644 (1992-10-01), Martin et al.
patent: 5261009 (1993-11-01), Bokser
patent: 5465309 (1995-11-01), Johnson
patent: 59-95682 (1984-06-01), None
“Contextual Word Recognition Using Probabilistic Relaxation Labeling”, by A. Goshtasby et al.,Pattern Recognition, vol. 21, No. 5, pp. 455-462, 1988.
“A Simplified Heuristic Version of a Recursive Bayes Algorithm for Using Context in Text Recognition”, by R. Shinghal et al., IEEE Transactions on Systems, Man, and Cybernetics, vol. SMC-8, No. 5, pp. 412-414, May 1978.
Seong-Whan Lee et al., “Efficient Postprocessing Algorithms for Error Correction in Handwritten Hangul Address and Human Name Recognition”,, Department of Computer Science, Chungbuk National University, IEEE 1993, pp. 232-235.
Katsumi Marukawa et al., A Post-Processing Method for Handwritten Kanji Name Recognition Using Furigana Information, Proceedings of the Second International Conference on Document Analysis and Recognition, IEEE Computer Society Press, Oct. 20-22, 1993, pp. 218-221.
Tsukumo, Jun “Handprinted Kanji Character Recognition Based on Flexible Template Matching” Pattern Rec. 1992 11thu th Int'l, vol. II, Conf. B.*
Sinha et al. “Visual Text Recognition through Contextual Processing” in Pattern Recognition vol. 21, No. 5 pp. 463-479, 1988.
ATR Auditory
Chang Jon
McDermott & Will & Emery
LandOfFree
Character recognition system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Character recognition system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Character recognition system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2499491