Optical character recognition system

• US 4,034,343 A
• Filed: 10/01/1976
• Issued: 07/05/1977
• Est. Priority Date: 10/01/1976
• Status: Expired due to Term

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Wilmer, Michael E.
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US05/728,666
• Time in Patent Office: 277 Days
• Field of Search: 340/146.3 MA, 340/146.3 H, 340/146.3 J, 340/146.3 R
• US Class Current: 382/295
• 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
• CPC Class Codes: G06V 30/196 using sequential comparisons of the image signals with a plurality of references

Abstract

An optical character recognition system for identifying typewritten characters and converting such to standard code for transmission. To achieve recognition of a character, the character, as it is scanned, is recorded in a corresponding array in memory and the memory address for each scan coordinate is then modified to correspond with the virtual character center. This coordinate center of the character is derived by inspection of the scanning array in parallel with the memory storing pass to determine the two adjacent all white rows and columns bounding the character. The coordinate center is then calculated by dividing the bounded row and column increment by two. The recognition of the character pattern is then performed by a set of Boolean discriminants selected to achieve the highest probability of in - character hits against a lowest probability of misses. Print fidelity, quantum misregistration and other character edge noise are filtered by selective OR-ing of test points along an axis having the highest incidence of noise.

Claims

1. A character recognition system adapted for use with a facsimile scanner for sampling the reflectivity of a document medium comprising:

memory means connected to said scanner for storing at corresponding address locations thereof the corresponding sampled reflectivity signals of said document medium;

isolation means operatively connected to said scanner for isolating in the output thereof the coordinates of a character on said document medium; and

pattern generating means connected to compare selected ones of said reflectivity signals in said memory means against a predetermined set of character patterns transposed by said character coordinates of said isolating means, said generating means including selected test points in said set of patterns combined in a logical OR to accommodate statistical variations in said isolated character.

2. Apparatus according to claim 1 wherein:

said set of character patterns includes a plurality of subsets each corresponding to a particular alphanumeric character, the test points in each subset being selected to substantially maximize hits on one isolated character and to substantially minimize hits on all other characters.

3. Apparatus according to claim 2 wherein:

said test points in one subset are compared sequentially against the contents of said memory means and on the first occurence of a non-comparison the next subset is initiated.

4. A character recognition system for automatically recognizing alphanumeric characters on a document medium, comprising:

document scanning means adapted to sense the elemental reflectivity of said document medium and producing a plurality of reflectivity signals indicative thereof;

memory means connected to said scanning means for storing said reflectivity signals in a matrix corresponding to coordinate locations of said document medium;

character isolation means connected to said scanning means for discriminating the vertical and horizontal boundary coordinates of a character on said document medium and storing a character center address corresponding thereto; and

recognition means adapted to contain a plurality of test patterns corresponding to a set of alphanumeric characters and connected to said memory and isolating means for transposing said test patterns by said center address and comparing said transposed test pattern against corresponding reflectivity signals in said memory means.

5. Apparatus according to claim 4 wherein:

said test pattern comprises a plurality of test point coordinates selected to produce a substantial maximum of hits on one character with a corresponding substantial minimum of hits on all other characters.

6. Apparatus according to claim 5 wherein:

selected ones of said test points are combined according to a logical OR.

7. Apparatus according to claim 6 wherein:

said scanning means includes serial sequencing means for arranging said reflectivity signals into a binary stream; and

said isolation means includes serial inspection means connected to receive said binary stream for inspecting selected sections of said stream and producing a first and second horizontal and a first and second vertical set of bounding signals according to the absence of character reflectivity in said selected sections.

8. Apparatus according to claim 7 wherein:

said test points of a test pattern are selected in sequence for transposition and comparison against a corresponding reflectivity signal as long as each comparison is successful.

9. In a pattern recognition system adapted for use in character recognition and including scanning means disposed for relative movement adjacent a document medium, said scanning means producing a serial stream of digital data indicative of the reflectivity of said medium at corresponding subjacent locations comprising:

11. Apparatus according to claim 9 wherein:

said recognition means further includes comparison means for comparing said test patterns at said second address coordinates with a corresponding first address coordinate in said memory means.

13. Apparatus according to claim 11 wherein:

said test points of one test pattern are compared in sequence against said memory means.

10. memory means operatively connected to receive said serial stream of digital data for recording thereof at corresponding addressable locations;

address generating means operatively connected to said scanning means for producing a first address code in correspondence with said document medium;

multiplexing means operatively connected to receive said first address code in a first state thereof and a second address code in a second state thereof for selecting said addressable locations in said memory means;

boundary discriminating means operatively connected to receive said serial stream in parallel with said memory means for discriminating the vertical and horizontal boundries of a character on said document medium and recording the boundary coordinate thereof;

recognition means adapted to contain selected pattern codes of a plurality of characters; and

combining means connected to receive said pattern codes and said boundary coordinates for providing said second address coordinates according to a predetermined combination thereof.

12. Apparatus according to claim 10 wherein:

said test points in each said test pattern are selected to substantially maximize hits on one character and to minimize hits on all other characters.