Method of processing character blocks with optical character reader

• US 5,033,098 A
• Filed: 08/27/1990
• Issued: 07/16/1991
• Est. Priority Date: 03/04/1987
• Status: Expired due to Term

• Current Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Original Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Inventors: Tanaka, Hideaki, Konya, Minehiro, Katsurada, Morihiro
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Mancuso, Joseph
• Application Number:
  US07/574,386
• Time in Patent Office: 323 Days
• Field of Search: 382/9, 382/48, 382/18, 382/40, 382/39, 382/57
• US Class Current: 382/171
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/153 using recognition of characters or words
• CPC Class Codes: G06V 30/268 Lexical context

Abstract

To efficiently process character blocks including mutually contacting characters by an optical character reader, not only are histograms prepared in a known manner but after candidate positions are determined from a histogram according to a specified set of criteria, their combinations are considered and the characters which are recognized corresponding to each combination of candidate positions are used to obtain average similarity values. Spellcheck routines and these average similarity values are considered together to identify characters which are contacting each other.

Claims

1. A method of processing character blocks by an optical alphanumeric character reader comprising the steps of

extracting a word image from a pictorial image,

extracting images of individual characters from said word image,

identifying each of said individual character images to obtain a character array,

carrying out a spellcheck of said obtained character array by comparing the obtained character array with words stored in a reference dictionary for judging whether or not the character array matches with said words in said reference dictionary,

assigning possibility numbers, if said obtained character array is not found as a word in a reference dictionary by said spellcheck or if any of said individual characters has an image longer than a predetermined maximum width, individually to each of characters and character blocks of said obtained character array in the order of the widths thereof,

obtaining a histogram from the image of each of said characters and character blocks, said histogram being taken in the perpendicular direction to the line of said character block,

obtaining from said histogram candidate positions where the frequency value is no larger than a predetermined value by excluding positions within a predetermined width from both ends of said character block and, if a plurality of said candidate positions are within a predetermined minimum width, considering only one of said plurality of said candidate positions having the smallest frequency value,

determining for each combination of said candidate positions a set of recognized characters and an average similarity value for said recognized characters,

forming combinations of said candidate positions sequentially in the order of said possibility numbers,

performing spellchecks on said combinations sequentially in the order determined by said average similarity values, and

selecting the first word found by said spellcheck.

2. The method of claim 1 wherein each of said candidate positions is selected where the frequency value is smaller than those at adjacent positions thereof.