Adaptive optical character recognition on a document with distorted characters

A computer program product stored on a non-transitory tangible computer readable storage medium for performing adaptive optical character recognition on a document with distorted characters, the computer program including code for:

• providing a group of images of candidate characters;
• receiving a document in an image format;
• segmenting the document so as to provide a sub-image of a segmented character;
• and
• for one or more of the candidate characters:
• performing a distortion-correction transformation on the segmented character, so as to modify the segmented character in the sub-image into a transformed character more closely resembling the candidate character;
• comparing the transformed segmented character to the candidate character so as to calculate a comparison score;
• analyzing the distortion-correction transformation so as to assign the transformation a distortion score;
• and
• determining whether to identify the segmented character with the candidate character, responsively to both the comparison score and the distortion score.

A data processing system comprising:

• a processor;
• a computer usable medium connected to the processor, wherein the computer usable medium contains a set of instructions for performing adaptive optical character recognition on a document with distorted characters, wherein the processor is designed to carry out a set of instructions to:
• provide a group of images of candidate characters;
• receive a document in an image format;
• segment the document so as to provide a sub-image of a segmented character;
• and
• for one or more of the candidate characters:
• perform a distortion-correction transformation on the segmented character, so as to modify the segmented character in the sub-image into a transformed character more closely resembling the candidate character;
• compare the transformed segmented character to the candidate character so as to calculate a comparison score;
• analyzing the distortion-correction transformation so as to assign the transformation a distortion score;
• and
• determining whether to identify the segmented character with the candidate character, responsively to both the comparison score and the distortion score.