Recognition method and system
First Claim
1. A method (30) of recognizing handwriting input, comprising the steps of:
- a) providing (32) a database of characters to be recognized, wherein characters are defined by a sequence of written reference strokes;
b) acquiring (34) written spatiotemporal input from a user in the form of discrete input strokes;
c) comparing (36) the discrete input strokes to the database of characters to obtain a comparison result; and
d) selecting and updating (38) a plurality of candidate character sequences based on (i) the comparison result and (ii) a candidate character sequence history derived from at least two sequential strokes previously acquired in step b, wherein the candidate character sequence history comprises a plurality of candidate characters.
1 Assignment
0 Petitions
Accused Products
Abstract
Techniques for recognizing discrete multi-component symbolic input from a user can be applied to, for example, handwriting or speech. The techniques can include providing a database of model input sequences, where each model input sequence corresponds to a symbol to be recognized. Input functions, for example, discrete strokes, are obtained from a user and segmented into a sequence of discrete components. Hypothesis symbol sequences are obtained by comparing the discrete components to a database of symbols to be recognized and updating hypothesis symbol sequences based on the results of the comparison and hypothesis symbol sequence history from input previously acquired in time.
-
Citations
32 Claims
-
1. A method (30) of recognizing handwriting input, comprising the steps of:
-
a) providing (32) a database of characters to be recognized, wherein characters are defined by a sequence of written reference strokes; b) acquiring (34) written spatiotemporal input from a user in the form of discrete input strokes; c) comparing (36) the discrete input strokes to the database of characters to obtain a comparison result; and d) selecting and updating (38) a plurality of candidate character sequences based on (i) the comparison result and (ii) a candidate character sequence history derived from at least two sequential strokes previously acquired in step b, wherein the candidate character sequence history comprises a plurality of candidate characters. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24)
-
-
25. A method of machine recognition of discrete multi-component symbolic input from a user, comprising:
-
a) providing a database of model input sequences, each model input sequence corresponding to a symbol to be recognized; b) acquiring an input function from a user, the input function including a time function; c) segmenting the input function into a sequence of input components; and d) determining a plurality of hypothesis symbol sequences, wherein at least one of the hypothesis symbol sequences is updated for each acquired input component based on a comparison of the input component to the database and based on a hypothesis symbol sequence history from at least two components previously in time. - View Dependent Claims (26, 27)
-
-
28. A handwriting recognition system (10) comprising:
-
a) a capture subsystem (12) configured to accept spatiotemporal input from a user and output a digitized representation of the spatiotemporal input; b) a processing subsystem (14) coupled to the capture subsystem to accept the digitized representation and determine recognized characters, the processing subsystem configured to; i) separate the spatiotemporal input into discrete input strokes; ii) compare the discrete input strokes to a database of model character stroke sequences to obtain a plurality of character scores; and iii) determine a plurality of candidate character sequences, wherein A) each candidate character sequence is updated to append a hypothesized character, B) a score for each candidate character sequence is updated based on a previous candidate character sequence score, where the previous candidate character sequence score is from a point previous in time corresponding to a number of strokes equal to the character length of the hypothesized character, and C) the score for each candidate character sequence is combined with a corresponding one of the character scores. - View Dependent Claims (29, 30, 31)
-
-
32. A non-transitory computer readable medium having computer readable program code embodied thereon for recognizing handwritten characters in a computer system having a user interface, comprising:
-
a) computer program code for providing a database of characters to be recognized, wherein for each character a character length sequence of reference strokes is defined; b) computer program code for acquiring spatiotemporal input from the user interface and outputting discrete input strokes; c) computer program code for comparing the discrete input strokes to the database to obtain a plurality of character scores; d) computer program code for determining a plurality of candidate character sequences, wherein the candidate character sequences are updated as the input strokes are acquired and at least one candidate character sequences is updated based on a candidate character sequence history from at least two strokes previously acquired in time; and e) computer program code for outputting a recognized character.
-
Specification