Method of optical character recognition

• US 4,731,861 A
• Filed: 08/26/1983
• Issued: 03/15/1988
• Est. Priority Date: 08/26/1983
• Status: Expired due to Term

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Petersen, Steven N., Helms, Ramon E., Blanton, Keith A.
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US06/527,155
• Time in Patent Office: 1,663 Days
• Field of Search: 382/9, 382/48
• US Class Current: 382/174
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/146 Aligning or centring of the image pick-up or image-field

Abstract

A method for scanning a page and locating individual characters on that page so that character recognition can be performed is disclosed. A page containing characters to be read is scanned with successive scan lines from top to bottom and the data contained in each scan line is transmitted to a buffer memory for temporary storage. An additional set of data which represents the significant amount of character information within a scan line is also generated and stored in a separate memory. When the microprocessor is ready to begin recognizing characters the set of data which indicates those scan lines containing significant character data is accessed and read. The microprocessor then accesses a different set of data which represents the significant character information as arranged in vertical pixel lines. When the microprocessor locates a vertical pixel line having significant character information the microprocessor then begins reading data from the buffer memory at locations as determined by the external data sets. The microprocessor does not read data in the buffer memory unless it corresponds to areas in which significant character data is located. The microprocessor is able to quickly scan an entire page and locate all characters within the page to provide for fast optical character recognition.

Claims

1. The method of isolating and reading character data for character recognition comprising:

(a) scanning a portion of a page having characters thereon with an imager to generate character data arranged in scan lines representative of said character and storing said character data,

(b) generating a first data set indicative of the amount of character data in each scan line,

(c) determining that a scan line contains significant character data only when corresponding data from said first set is above a scan line threshold value,

(d) reading character data only in scan lines which contain significant character information.

2. The method according to claim 1 further comprising:

(a) generating a second data set indicative of the character content of pixel lines perpendicular to said scan lines,

(b) determining that a perdendicular pixel line contains significant character data only when corresponding data from said second set is above a pixel line threshold value,

(c) reading character data only from pixel lines which contain significant character data.

3. The method according to claim 2 further comprising:

(a) determining the beginning and ending scan lines of a text line at characters,

(b) determining the beginning and ending perpendicular pixel lines for a character within said text line,

(c) reading said character data beginning at a point between said beginning and ending perpendicular pixel lines,

(d) generating parameter data from said reading sequence,

(e) recognizing said character from said parameter data and providing a number representative of said character in an output buffer,

(f) repeating steps b-e until all characters within said text line have been read,

(g) outputting said numbers from said output buffer when an entire text line has been read,

(h) repeating steps a-f until all text lines on said page have been read.

4. The method accroding to claim 3 wherein characters adjacent to any "O" or "zero" in said output buffer are read and when either adjacent character is alphabetic setting any "zero" to an "O", otherwise setting an "O" to a "zero" prior to outputting said numbers from said output buffer.

5. The method according to claim 3 wherein the parameter data is comprised of the height, width, perimeter and area within the periphery of said character.

6. The method according to claim 5 wherein determining the beginning and ending perpendicular pixel lines for a character is comprised of:

(a) reading data from said third set in sequential order,

(b) selecting as the beginning of a character the left most pixel line whose corresponding data from the second set is above a threshold value,

(c) selecting as the end of a character the right most pixel line whose corresponding data from the second set is above the threshold value when the right adjacent pixel line is below the threshold value.

7. The method according to claim 6 wherein a character is determined to be too wide to be a single character when the number of pixel lines from the beginning pixel line to the end pixel line exceeds a single character threshold value.

8. The emthod according to claim 7 wherein a character which is too wide to be a single character is split into two characters by selecting as the right most pixel line of said character a pixel line which is a predetermined number of pixel lines from the left most pixel line.

9. THe method according to claim 6 wherein the character is rejected as not a character when the area is less than a threshold area value.

10. The method according to claim 6 comprising reading character data immediately above said character within said text line when the perimeter of said character is below a threshold perimeter value indicating that a two-part character might exist.

11. The method according to claim 3 further comprising:

placing the appropriate spaces between sequential characters by determining the number of perpendicular pixel lines between said characters and placing the appropriate space in said output buffer.

12. The method according to claim 3 further comprising:

(a) determining if adjacent apostrophies are in said output buffer,

(b) determining the number of pixel lines between any adjacent apostrophies,

(c) changing the numbers representing the two apostrophies to a single number representing a quotation mark when the number of pixels between them is less than an apostrophe threshold value.

13. The method of isolating a text line of characters comprising:

(a) scanning a portion of a page having characters thereon with an imager to generate character data arranged in scan lines representative of said character and storing said character data,

(b) generating a first data set indicative of the amound of character data in each scan line,

(c) reading data from said first set,

(d) selecting as the beginning of a text line the upper most scan lines whose corresponding data from the first set is above a scan line threshold value,

(e) selecting as the end of a text line the lower most scan line whose corresponding data from the first set is above the scan line threshold value when the lower adjacent scan line is below the scan line threshold value.

14. The method according to claim 13 wherein individual characters within said text line are isolated by the method comprising:

(a) generating a second data set indicative of the amount of character information in pixel lines which are perpendicular to the scan lines.

(b) reading data from said second set,

(c) selecting as the beginning of a character the left most scan line whose correpsonding data from the second set is above the pixel line and threshold value,

(d) selecting as the end of a character the right most pixel line whose corresponding data from the second set is above the pixel line threshold value from the right adjacent pixel line is below the pixel line threshold value.

15. The method of isolating text lines for character recognition comprising:

(a) scanning a portion of a page having characters thereon with an imager to generate character data arranged in scan lines representative of said character,

(b) storing said character data in a plurality of storage means,

(c) determining the upper most scan line of a text line of characters,

(d) determining the lower most scan line of the text line of characters,

(e) storing all character data within the isolated text line in a single storing means,

(f) storing the subsequent text line in a separate storing means.