Recognition method and system

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.

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.

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.