Automatic handwriting recognition using both static and dynamic parameters

• US 5,544,264 A
• Filed: 05/25/1995
• Issued: 08/06/1996
• Est. Priority Date: 01/27/1993
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bellegarda, Jerome R., Nahamoo, David, Nathan, Krishna S.
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Tran, Phuoc
• Application Number:
  US08/451,001
• Time in Patent Office: 439 Days
• Field of Search: 382/224, 382/225, 382/226, 382/227, 382/228, 382/159, 382/160, 382/161, 382/186, 382/187, 382/188, 382/189, 382/215, 382/190, 382/195, 382/203, 382/120, 382/122, 382/123
• US Class Current: 382/188
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition
• CPC Class Codes: G06V 30/1918 Fusion techniques, i.e. combining data from various sources, e.g. sensor fusion
• CPC Class Codes: G06V 30/347 Sampling; Contour coding; Stroke extraction
• CPC Class Codes: G06V 30/373 using a special pattern or subpattern alphabet

Abstract

Methods and apparatus are disclosed for recognizing handwritten characters in response to an input signal from a handwriting transducer. A feature extraction and reduction procedure is disclosed that relies on static or shape information, wherein the temporal order in which points are captured by an electronic tablet may be disregarded. A method of the invention generates and processes the tablet data with three independent sets of feature vectors which encode the shape information of the input character information. These feature vectors include horizontal (x-axis) and vertical (y-axis) slices of a bit-mapped image of the input character data, and an additional feature vector to encode an absolute y-axis displacement from a baseline of the bit-mapped image. It is shown that the recognition errors that result from the spatial or static processing are quite different from those resulting from temporal or dynamic processing. Furthermore, it is shown that these differences complement one another. As a result, a combination of these two sources of feature vector information provides a substantial reduction in an overall recognition error rate. Methods to combine probability scores from dynamic and the static character models are also disclosed.

Claims

1. A method for operating a handwriting recognition system, comprising the steps of:

receiving, from a handwriting transducer means, x-y coordinate information generated by a writer;

extracting temporally-based feature vectors from the x-y coordinate information;

extracting spatially-based feature vectors from the x-y coordinate information; and

identifying a most probable character (U) from a set of possible characters (a) as being a character that is written by the writer, the step of identifying employing a probabilistic method that considers both the temporally-based feature vectors and the spatially-based feature vectors in accordance with the expression

space="preserve" listing-type="equation">Pr(U|a.sub.j)=Pr(U.sub.t |a.sub.j).sup.1/2 Pr(U.sub.d |a.sub.j)1/2,

where Pr(U_d |a_j) is determined from the spatially-based feature vectors and where Pr(U_t |a_j) is determined from the temporally-based feature vectors.

2. A method for operating a handwriting recognition system to determine if a character written by a writer is a character within a character alphabet (a), comprising the steps of:

receiving, from a handwriting transducer means, x-y coordinate information generated by the writer;

partitioning the x-y coordinate information into a plurality of frames w_t;

decoding, using temporally-based feature vectors and temporally-based character models, all frames w_t to obtain a first probability Pr(w_t |a_j) ∀

j, ∀

t;

decoding, using spatially-based feature vectors and spatially-based character models, the x-y coordinate information associated with a spatial image block s to obtain a second probability Pr(s_d |a_j) ∀

j;

for each image block s, identifying all of the frames w_t that lie within that image block;

defining a third probability as, ##EQU11## where T is the number of frames of w_t within s; and

determining a fourth probability that an image block s, given a_j, is a most-probable character that is written by the writer in accordance with,

space="preserve" listing-type="equation">Pr(s|a.sub.j)=Pr(s.sub.t |a.sub.j).sup.α

Pr(s.sub.d |a.sub.j).sup.1-α

.

3. A handwriting recognition system, comprising:

a handwriting transducer means that outputs x-y coordinate information generated by a writer;

means, coupled to an output of said transducer means, for extracting temporally-based feature vectors from the x-y coordinate information;

means, coupled to said output of said transducer means, for extracting spatially-based feature vectors from the x-y coordinate information; and

means, coupled to both said extracting means, for identifying a most probable character (U) from a set of possible characters (a) as being a character that is written by the writer, said means for identifying operating in accordance with a probabilistic method that considers both the temporally-based feature vectors and the spatially-based feature vectors in accordance with the expression

space="preserve" listing-type="equation">Pr(U|a.sub.j)=Pr(U.sub.t |a.sub.j).sup.1/2 Pr(U.sub.d |a.sub.j).sup.1/2,

Pr(U_d |a_j) is determined from the spatially-based feature vectors and where Pr(U_t |a_j) is determined from the temporally-based feature vectors.