Apparatus and method for recognizing characters

• US 4,180,799 A
• Filed: 04/21/1978
• Issued: 12/25/1979
• Est. Priority Date: 04/21/1978
• Status: Expired due to Term

• Current Assignee: Caere Corp. (Mexico) (Microsoft Corporation)
• Original Assignee: Caere Corp. (Mexico) (Microsoft Corporation)
• Inventors: Smith, William R.
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US05/898,822
• Time in Patent Office: 613 Days
• Field of Search: 340/146.3 H, 340/146.3 AC, 340/146.3 AG, 340/146.3 MA, 340/146.3 J, 340/146.3 R
• US Class Current: 382/196
• CPC Class Codes: G06V 10/20 Image preprocessing
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 10/40 Extraction of image or video features
• CPC Class Codes: G06V 10/421 by analysing segments intersecting the pattern

Abstract

An optical reader for recognizing printed characters, such as alpha-numeric characters, is disclosed. The characters are scanned in parallel, vertical slices by a photodiode array contained in a hand-held wand which is manually moved over the printed characters. The resultant video signals are examined for predetermined features, such as gaps, bars, strokes, etc. These features are encoded into a digital word for each slice. A logic tree analysis is used in which each new digital word is compared to words along the tree to direct the analysis to branches or substances. The continued comparison leads to a positive recognition of a single character. The raw video is not stored as in prior art systems, but rather the video signals are processed in a serial manner with feature identification occurring without storage. The processing circuitry thus is efficiently used since the video signals are processed as they occur.

Claims

1. An apparatus for recognizing characters such as alpha-numeric characters comprising:

scanning means for viewing a character and for providing video signals representative of generally parallel slices of such character;

feature extraction means, coupled to receive said video signals from said scanning means, for detecting features from said signals as they are received and for assigning predetermined codes to said detected features, said feature extraction means providing a sequence of coded signals representing said detected features for said generally parallel slices of such character;

logic tree means, coupled to receive said coded signals from said feature extraction means each time a change occurs in said coded signals, for logically evaluating each of said coded signals to determine the possible characters having features corresponding to said coded signals, such tree means narrowing such possible characters as said coded signals are evaluated until such character is identified;

whereby characters are identified without storage of video data.

2. The apparatus defined by claim 1 wherein said scanning means includes a dynamic thresholding means for providing a dynamic threshold potential which is compared with said video signal to compensate for variation in printing density and background coloring.

3. The apparatus defined by claim 2 wherein said scanning means includes a light-sensitive linear array which is held in general vertical alignment with such character during said viewing with relative motion occurring between said array and such character to provide said generally parallel slices.

4. The apparatus defined by claim 1, including a plurality of said linear arrays aligned parallel to one another coupled to a weighting network for combining signals from said arrays to provide a channel of said video signals.

5. The apparatus defined by claim 4 including two of said channels of said video signals.

6. The apparatus defined by claim 5 including a zone control means for compensating for vertical misalignments between the field-of-view of said scanning means and viewed characters.

7. The apparatus defined by claim 6 wherein said zone control means provides a zone control signal which is used to select a portion of said video signals for said detection.

8. The apparatus defined by claims 6 or 7 wherein said zone control means provides an updated zone control signal after scanning a character to provide said compensation for misalignments for the next character.

9. The apparatus defined by claim 1 wherein said sequence of coded signals provided by said feature extraction means comprises a plurality of digital words, one of said words being provided for each of said slices.

10. The apparatus defined by claim 9 including a first-in, first-out register coupled to receive said digital words and coupled to communicate said digital words to said logic tree means.

11. The apparatus defined by claim 10 including storage means for storing each of said digital words in said register, a digital number representing the number of said slices during which said word was detected.

12. An apparatus for recognizing characters, such as alpha-numeric characters, comprising:

scanning means for viewing a character and for providing a first channel and a second channel of video siganls representing generally parallel slices through such character;

direction detection means for determining the direction of relative movement between said scanning means and such character, said detecting means coupled to receive said first channel and second channel of video signals and for making said direction determination based on said video signals;

first feature extracting means for receiving at least one of said first and second channels of video signals and for detecting first predetermined features;

second feature extracting means for receiving said first and said second channels of video signals and for detecting second predetermined features;

encoding means for encoding digital words with said first and said second predetermined features as said features are detected by said first and said second feature extracting means; and

logic tree means coupled to receive said digital words from said encoding means and coupled to said direction detection means, for analyzing said words to determine possible characters possessing said detected features;

whereby characters may be recognized with storage of video data.

13. The apparatus defined by claim 12 wherein said first predetermined features include the detection of bars and gaps.

14. The apparatus defined by claim 13 wherein said second predetermined features include an indication of the direction of character development.

15. A method for recognizing characters comprising the steps of:

generating a plurality of video signals, each representing sequential parallel slices of a character;

examining each of said video signals to identify predetermined features of characters;

encoding digital words with identified ones of said predetermined features for each of said slices;

examining said digital words to determine a change between one word and the next to provide a sequence of changed ones of said digital words;

analyzing said sequence of changed words, one after the other, to determine the possible characters possessing said identified features, said analysis of said changed digital words continuing until a character is recognized;

whereby a character is recognized with a minimum amount of memory.

16. The method defined by claim 15 including counting the number of slices between said changed words to provide width indication of said identified features.

17. The method defined by claim 15 wherein one of said predetermined features is a bar in a character.

18. The method defined by claim 17 wherein another of said predetermined features is a stroke which is shorter than said bar.

19. The method defined by claim 18 wherein another of said predetermined features is a gap.

20. In an apparatus for recognizing characters, such as alpha-numeric characters, a means for determining the direction of character development comprising:

scanning means for viewing a character and for providing first and second video signals representing two parallel, spaced-apart slices through such character;

first feature extracting means for receiving said first and second video signals and for detecting first predetermined features from said first and second video signals;

second feature extracting means for receiving signals representing said detected first predetermined features for said first and second video signals and for detecting second predetermined features based on a common time comparison of said first predetermined features;

whereby said second predetermined features provide an indication of the direction of character development.