Holistic-analytical recognition of handwritten text

A computer storage medium readable by a computing system and encoding a computer program of instructions for executing a computer process for recognizing handwritten cursive text, the computer process comprising:

• loading into the computing system an image of an input phrase of the cursive text;
• identifying features of the input phrase, each feature representing at least a portion of a character in the input phrase;
• matching features of the input phrase as a whole against features of a plurality of reference phrases from a stored lexicon and generating a holistic answer list consisting of as answers reference phrases from the plurality of reference phrases that are most similar to the input phrase;
• constructing a character segmented features list from the features of the input phrase and based on answers in the holistic answer list, the character segmented features list being a list of character features segmented into sets by character for each segmentation hypothesis used for each answer in the holistic answer list;
• translating the image of the input phrase into character images for each character position in the input phrase based upon the character segmented features list;
• analytically recognizing character images in each character position as translated from the image of the input phrase and generating an analytical answer list consisting of analytical answer phrases, each analytical answer phrase being a string of characters individually recognized from the character images making up the input phrase;
• and
• comparing the holistic answer list and the analytical answer list to determine a best recognition answer for the input phrase.

A system for recognizing handwritten cursive text, comprising:

• at least one processing unit;
• a memory, operatively connected to the at least one processing unit and including instructions that, when executed by the at least one processing unit, cause the at least one processing unit to perform a method comprising:
• loading into the computing system an image of an input phrase of the cursive text;
• identifying features of the input phrase, each feature representing at least a portion of a character in the input phrase;
• matching features of the input phrase as a whole against features of a plurality of reference phrases from a stored lexicon and generating a holistic answer list consisting of as answers reference phrases from the plurality of reference phrases that are most similar to the input phrase;
• constructing a character segmented features list from the features of the input phrase and based on answers in the holistic answer list, the character segmented features list being a list of character features segmented into sets by character for each segmentation hypothesis used for each answer in the holistic answer list;
• translating the image of the input phrase into character images for each character position in the input phrase based upon the character segmented features list;
• analytically recognizing character images in each character position as translated from the image of the input phrase and generating an analytical answer list consisting of analytical answer phrases, each analytical answer phrase being a string of characters individually recognized from the character images making up the input phrase;
• and
• comparing the holistic answer list and the analytical answer list to determine a best recognition answer for the input phrase.