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.
312 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