Optical character recognition apparatus

• US 4,747,149 A
• Filed: 03/17/1987
• Issued: 05/24/1988
• Est. Priority Date: 03/17/1986
• Status: Expired due to Fees

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Umeda, Tetsuo, Kasuya, Shigenobu
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US07/026,890
• Time in Patent Office: 434 Days
• Field of Search: 382/18, 382/51, 382/54, 358/282, 358/284
• US Class Current: 382/194
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/162 Quantising the image signal
• CPC Class Codes: H04N 1/403 Discrimination between the two tones in the picture signal of a two-tone original

Abstract

The disclosure describes a character recognition apparatus comprising a scanner for optically scanning a prescribed area and outputting a scanned signal, a binary processing circuit for receiving the scanned signal and producing a plurality of binary signals, a counting mechanism for counting black meshes in the scanned area, a counting mechanism for counting insulated meshes, a mechanism for calculating the ratio of the number of black meshes and the number of insulated meshes, and a mechanism for selecting one optimum binary signal from the plurality of binary signals on the basis of a comparison of the ratios.

Claims

1. An optical character recognition apparatus comprising:

means for optically scanning a prescribed scanning area to produce a scanned signal corresponding to a character contained in said prescribed scanning area;

means responsive to said scanned signal for producing a plurality of binary signals each having a different characteristic, each binary signal being quantized to construct said prescribed scanning area of a plurality of meshes arranged to form a matrix and being represented by black meshes each indicating a part of said character and by white meshes each indicating a background;

means for counting black meshes contained in said prescribed scanning area with respect to said plurality of binary signals to deliver a plurality of black mesh numbers corresponding to said plurality of binary signals, respectively;

means for counting insulated meshes with respect to said plurality of binary signals to deliver a plurality of noise mesh numbers corresponding to said plurality of binary signals, said insulated mesh being a white mesh surrounded by black meshes and a black mesh surrounded by white meshes;

means for calculating a plurality of ratios between said plurality of black mesh numbers and said plurality of noise mesh numbers, said plurality of ratios corresponding to said plurality of binary signals, respectively;

means for determining an optimum binary signal from said plurality of binary signals on the basis of comparison of said plurality of ratios calculated by said calculating means; and

means for recognizing said character contained in said prescribed scanning area on the basis of said optimum binary signal determined by said determining means.

2. An optical character recognition apparatus as claimed in claim 1, wherein said determining means includes:

means for storing said plurality of binary signals,

means for comparing said plurality of ratios with each other to produce a selection signal to select said optimum binary signal, and

means for reading out said optimum binary signal from said storing means in response to said selection signal.

3. A character recognition apparatus comprising:

means for scanning a predetermined scanning area to produce a scanned binary signal corresponding to a character contained in said predetermined scanning area;

means for converting said scanned binary signal into a digital signal, said digital signal being constructed by a plurality of bits;

means for producing a plurality of binary signals from said digital signal by selecting a plurality of bits constructing said digital signal, each binary signal being quantized to include a plurality of meshes arranged to form a matrix corresponding to said predetermined scanning area;

means for counting black meshes each showing a part of a character contained in said predetermined scanning area with respect to said plurality of binary signals to deliver a plurality of black mesh numbers corresponding to said plurality of binary signals, respectively;

means for counting insulated meshes contained in said predetermined scanning area with respect to said plurality of binary signals to deliver a plurality of insulated mesh numbers corresponding to said plurality of binary signals, said insulated mesh being a black mesh surrounded by white meshes and a white mesh surrounded by black meshes, said white mesh showing other than said part of said character;

means for calculating a plurality of ratios between said plurality of black mesh numbers and said plurality of insulated mesh numbers, respectively;

means for comparing said plurality of ratios with each other to select one binary signal from said plurality of binary signals in accordance with the comparison result; and

means for recognizing said character contained in said predetermined scanning area in response to said one binary signal delivered from said comparing means.

4. A character recognition apparatus comprising:

means for scanning a character on a mail article with a predetermined scanning area to produce a scanned signal;

means coupled to said scanning means for producing a plurality of binary signals from said scanned signal, said plurality of binary signals being binary-coded under respective different conditions, each of said binary signals being quantized to construct said predetermined scanning area of a plurality of meshes arranged in a matrix form and being represented by black meshes each indicating a part of a character and by white meshes other than said black mesh;

means for recognizing said character on said mail article on the basis of said plurality of binary signals to produce a plurality of recognition results corresponding to said plurality of binary signals, respectively;

means for counting said black meshes contained said predetermined scanning area with respect to said pluraltiy of binary signals to output a plurality of the numbers of black meshes corresponding to said plurality of binary signals, respectively;

means for counting noise meshes contained said predetermined scanning area with respect to said plurality of binary signals to output a plurality of the numbers of noise meshes corresponding to said plurality of binary signal, respectively, said noise mesh being a white mesh surrounded by black meshes and a black mesh surrounded by white meshes;

means for calculating a plurality of ratios between said plurality of the numbers of black meshes and said plurality of the numbers of noise meshes with respect to said plurality of binary signals;

means for selecting an optimum binary signal from said plurality of binary signals on the basis of said plurality of ratios calculated by said calculating means to deliver a selection signal corresponding to said optimum binary signal; and

means responsive to said plurality of recognition results delivered from said recognizing means and said selection signal delivered from said selection means for determining a recognition output corresponding to said character.

5. A character recognition apparatus as claimed in claim 4, wherein said determining means determines said recognition output by selecting one from said plurality of recognition results in accordance with said selection signal when said plurality of recognition results are not identical with each other, said selected one corresponding to said optimum binary signal.