Recognition system for run-on handwritten characters

• US 4,731,857 A
• Filed: 11/03/1986
• Issued: 03/15/1988
• Est. Priority Date: 06/29/1984
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Tappert, Charles C.
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US06/926,837
• Time in Patent Office: 498 Days
• Field of Search: 382/9, 382/13, 382/37, 382/38, 382/40, 382/24
• US Class Current: 382/178
• CPC Class Codes: G06V 10/10 Image acquisition document image scanning and transmission H04N1/00; control of digital cameras H04N23/60
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/153 using recognition of characters or words

Abstract

A method of processing a word with the segmentation and recognition steps combined into an overall scheme. This is accomplished by a three step procedure. First, potential or trail segmentation points are derived. This is done in a manner so as to ensure that essentially all true segmentation points are present but also includes extra or not true segmentation points. Second, all combinations of the segments that could reasonably be a character are sent to a character recognizer to obtain ranked choices and corresponding scores. Finally, the recognition results are sorted and combined so that the character sequences having the best cummulative scores are obtained as the best word choices. For a particular word choice there is a corresponding character segmentation, simply the segment combinations that resulted in the chosen characters. With this recognition scheme the initial character segmentation is not final and need not be highly accurate, but is subject to a lesser constraint of containing the true segmentation points.

Claims

1. A method of recognizing handwritten words, a word being formed by a plurality of characters that may touch or overlap each other, each character being formed by one or more strokes comprising the steps of:

(a) initializing an index of a current segmentation point of a character;

(b) considering a stroke of a handwritten character having a segmentation point and incrementing said index or terminating said method when no further strokes exist to be processed.

(c) examining recognitions on stroke sequences ending at the segmentation point corresponding to said index;

(d) grading the results of said examination of recognitions to rank potential character choices; and

(e) repeating steps (b) through (d) until all strokes for characters comprising a word have been processed.

2. A recognition method of claim 1 wherein said step of examining said recognitions comprises the steps of:

(cⁱ) determining the maximum number of strokes of a character in the alphabet to be recognized;

(cⁱⁱ) setting a register of the number of strokes to be examined by a recognizer;

(cⁱⁱⁱ) comparing the contents of said register with the maximum number of strokes and terminating the examination of recognitions when said register exceeds the maximum number of strokes;

(c^iv) when said register contents are less than or equal to the maximum number of strokes, performing examination of recognitions on a sequence of strokes corresponding to the contents of said register and ending in the stroke corresponding to the stroke of the segmentation point corresponding to the contents of said index;

(c^v) storing candidate characters and scores obtained during examination of recognitions; and

(c^vi) incrementing said register of the number of strokes to be examined; and

repeating steps (c¹¹¹) through (c^vi) until all stroke sequences up to the maximum number of strokes have been examined.

3. A recognition method of claim 2 wherein the step of grading the results of said examination of recognitions comprises the steps of:

setting a counter of the number of strokes sent to the recognizer;

comparing the contents of said counter with the maximum number of strokes and terminating the grading when said counter exceeds the maximum number of strokes; and

when said counter contents are less than or equal to the maximum number of strokes, assembling a list recognition choices and ranking by recognition choice scores corresponding to graded results.

4. A recognition method as in claim 3 wherein the step of assembling a list of recognition choices comprises initializing a choice index,

incrementing said choice index,

determining whether the contents of the choice index exceed a maximum number of choices to be listed and whether the recognition choice score exceeds a reject threshold;

if the recognition choice score exceeds said reject threshold and the maximum number of choices has not been exceeded creating a new recognition choice and inserting it into the list by increasing score and returning to the step of incrementing said choice index; and

if the recognition choice score does not exceed said reject threshold and the maximum number of choices has been exceeded, processing a next sequence of strokes for grading.

5. A recognition method as in claim 3 further comprising the step of comparing the recognition choices with a list of acceptable character strings;

and eliminating recognition choices having no corresponding acceptable character string.