Optical character isolation system, apparatus and method

• US 4,741,045 A
• Filed: 09/23/1983
• Issued: 04/26/1988
• Est. Priority Date: 09/23/1983
• Status: Expired due to Fees

• Current Assignee: FOOG Incorporated
• Original Assignee: Dest Corp.
• Inventors: Denning, Bruce S.
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Mancuso, Joseph
• Application Number:
  US06/535,410
• Time in Patent Office: 1,677 Days
• Field of Search: 382/9, 382/57, 382/48, 382/32, 382/3, 382/13, 382/40
• US Class Current: 382/178
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/155 Removing patterns interfering with the pattern to be recognised, such as ruled lines or underlines

Abstract

Low cost, high-speed, optical character isolation and page reconstruction system, method and apparatus are presented which overcome problems caused by copied pages, noise, underlines, skewed and bowed text, forms features, logos and signatures. As characters or noise are isolated and recognized, their corresponding bit patterns in memory are deleted. Recognized characters are isolated within entire words at a time to form page image records which are then linked to form lines of words. The text on the original page is then reconstructed from lines of words to yield output signals suitable for input to a host word processor.

Claims

1. Method of producing output data representative of isolated information markings within a class of such markings oriented on a document, the metod comprising the steps of:

vertically sensing the information markings in a selected sequence to produce word data manifestations of the sensed information markings on a word by word basis where said information markings are in the form of a word and where said class of markings are in the form of text,

storing the word data manifestations in substantially similar orientation to the orientation of the sensed information markings on the document,

selectively addressing stored word data manifestations oriented with respect to a selected reference location,

producing encoded output in recognition of selectively addressed stored word data manifestations as specific ones of the information markings within the class of such markings,

deleting from storage the selectively addressed word data manifestations which are recognized, and

(a) selectively addressing any remaining stored word data manifestations which are selectively oriented in storage with respect to the locations of the deleted word data manifestations, and

(b) successively producing encoded outputs in recognition of any successively addressed word data manifestations, and

(c) successively deleting from storage any successively addressed word data manifestations which are recognized, until such recognitions and deletions from storage of word data manifestations extend to a selected orientation limit,

and

selectively accumulating the encoded outputs to provide isolated output data representative of the information word markings and the orientations thereof on the document.

2. The method according to claim 1 wherein in the step of selectively addressing stored data manifestations, the selected reference location is the top-most stored data manifestation located maximally displaced to one side.

3. The method according to claim 2 wherein the step of successively addressing any remaining stored data manifestations continues to one limit of a contiguous array of data manifestations representing a continuous succession of sensed information markings on the document.

4. The method according to claim 3 comprising the step of establishing said one limit at a separation between contiguous data manfestations located not greater than a predetermined number of data manifestations displaced away from said selected reference location.

5. The method according to claim 3 comprising the step of successively addressing any remaining stored data manifestations to an opposite limit from said one limit of the contiguous array of data manifestations to establish an image record of said encoded outputs representative of a set of contiguous information markings from the beginning to the end thereof.

6. The method according to claim 5 comprising the steps of:

selectively addressing any remaining stored data manifestations commencing at the reference location at which the top-most remaining stored data manifestation maximally displaced to one side is located; and

successively addressing the stored data manifestations between opposite ends of a contiguous array thereof to establish an image record of said encoded outputs representative of another set of contiguous information markings on the document.

7. The method according to claim 6 comprising the step of:

coding the image records with data indicative of the orientations thereof within an array of stored data manifestations; and

wherein said step of selectively accumulating the encoded outputs includes

assembling the image records of encoded outputs according to selected coding data associated therewith to produce said output data representative of the information mailings and the orientations thereof on the document.

8. The method according to claim 1 wherein the step of producing encoded outputs includes:

forming image records, each represenative of a contiguous set of information markings on the document between left and right limits thereof, and each including coding data indicative of the location of the corresponding information markings on the document;

and including the step of assembling a plural number of image records in sequence according to coding data associated therewith to provide said output data representative of the information markings and the orientations thereof on the document.

9. Method according to claim 8 wherein each image record includes a representative baseline associated wtih the number of encoded outputs contained therein;

and comprising the step of;

identifying the image record in a plural number thereof having associated coding data representing a reference orientation; and

successively combining remaining ones of the plurality of image records in selected sequence according to coding data associated therewith and with the baselines thereof substantially aligned for reproducing the sequence of information markings on the documents, independently of any horizontal skew orientation thereof.

10. The method according to claim 9 wherein in the step of successively combining,

a first set of a number of image records combined in sequence represent a first line of information markings on the document;

and

a second set of a number of image records combined in sequence represent a second line of information markings displaced vertically on the document from the first line, and including the step of comparing coding data associated with the initial image record in each of the first and second sets thereof; and

altering coding data associated with the location of the initial image record in one of the first and second sets thereof if the difference between location coding data therefor is less than a preselected value for reproducing the first and second sets of image records with the beginnings thereof oriented in substantial vertical alignment, independently of any vertical skew orientation of the beginnings of the sets of corresponding information markings on the document.

11. Method according to claim 1 comprising the additional step of:

determining the magnitude of the addressed stored data manifestation within substantially contiguous boundaries;

deleting from storage as representing noise the addressed stored data manifestations which are determined to be less than a preselected magnitude;

processing for recognition the addressed stored data manifestations which are determined to be greater than the preselected magnitude.

12. Method according to claim 11 comprising the step of:

storing for reformation the data manifestations which are determined to be greater than the preselected magnitude and which do not provide encoded output in recognition thereof.

13. Method according to claim 1 for producing output data representative of information markings including underlines of the markings oriented on a document, the method comprising the step of:

isolating for recognition any separated segments of addressed stored data manifestations substantially at the locations of interruptions in horizontal lines of data manifestations being substantially vertically aligned.

14. Method according to claim 13 comprising the steps of:

determining the length of a contiguous set of addressed stored data manifestations between said interruptions;

deleting from storage the data manifestations representing underline between the interruptions for contiguous sets greater than a selected length; and

accumulating encoded outputs, including an encoded output representing underline, to provide output data representative of the information markings, and underlines thereof, and the respective orientations thereof on the document.

15. Method according to claim 14 comprising the steps of:

determining the length between interruptions of a contiguous set of addressed stored data manifestations remaining after delection of the data manifestations representing underline, where such remaining set has length greater than the normal width represented by one information marking, to identify information markings which touch or overlap; and

separating such remaining set at a selected location spaced from an end thereof that substantially equals the width represented by one information marking; and

buffer storing the data manifestations forming one of the separated segments of the set and deleting the same from storage, and alternatingly therewith;

producing encoded coding in recognition of the data manifestations forming the other of the separated segments of the set.

16. Method according to claim 15 wherein in the step of separating:

such remaining set is separated substantially at a location closest to said selected location at which the number of data manifestations forming the contiguous set is lowest.

17. Method according to claim 1 wherein:

the encoded outputs produced in recognition of selected addressed data manifestations include an encoded output indicating which information marking of the class, with baseline data therefor, is most likely represented by such data manifestations, and also an encoded output indicating which information marking of the class, with baseline data therefor, is next most likely represented by such data manifestations; and

the step of selectively accumulating the encoded outputs includes;

forming sets of encoded outputs representing contiguous information markings on the document by combining only those encoding outputs for the most likely and next most likely representations for which the associated baseline data is also in closest correlation.

18. Isolation apparatus for producing output signals indicative of information markings within a class of such markings oriented on a document, the apparatus comprising:

transducer means disposed with respect to a document for vertically producing word data signals representative of information markings at selected locations on the document on a word by word basis where said information markings are in the form of words and where said class of markings are in the form of text,

storge means coupled to the transducer means for storing word data signals therein in substantially similar orientation to the orientation of the corresponding information markings on the document,

circuit means coupled to the storage means for addressing word data signals stored therein at selected locations,

recognition circuit means responsive to the addressed data signals in the storage means for producing code signals indicative of the addressed word data signals corresponding substantially to an information word marking within the class of such markings,

means coupled to the storage means for removing from storage therein the stored word data signals corresponding to the information marking for which code signals were produced,

said circuit means being operative thereafter for addressing any remaining word data signals stored in said storage means at selected locations relative to the locations of previously removed word data signals,

said recognition circuit means producing isolation code signals indicative of such remaining addressed word data signals corresponding substantially to information marking within the class of such markings,

said means operating to remove from storage the word data signals for which code signals were produced until the removals of word data signals in storage extend to a selected orientation limit, and

compiling means coupled to receive said isolation code signals for assembling same in selected sequence to produce isolated output signals representative of the information markings and the orientations thereof on the document.

19. Document-handling system for producing isolation output signals representative of information markings oriented on the document, the system comprising:

sensing means mounted with respect to a document for vertically producing word data signals in response to detection of selected portions of information markings on the document on a word by word basis where said information markings are in the form of words and where said class of markings are in the form of text,

transport means coupled to the document for selectively displacing the document incrementally along a direction relative to the sensing means,

storage means coupled to the sensing means for storing word data signals at addressable locations therein,

circuit means coupled to the storage means for selectively addressing word data signals therein to produce isolation output signals representative of substantially corresponding information markings, and to remove from said storage means the selectively addressed word data signals for which isolated output signals are produced, and

control means coupled to said transport means and said storage means and said circuit means for successively displacing the document incrementally at a rate sufficient to maintain the storage means substantially filled with word data signals as the same are selectively addressed and removed therefrom.