System and method for recognizing visual indicia

• US 5,703,969 A
• Filed: 02/16/1996
• Issued: 12/30/1997
• Est. Priority Date: 01/21/1994
• Status: Expired due to Term

• Current Assignee: Martek, Inc.
• Original Assignee: Electroglas, Inc., Texas Instruments, Inc., Texas Tech University (Texas Tech University System)
• Inventors: Lovelace, Jerome R., Hastings, Howard V. III, Chang, Ning San, Hennessey, A. Kathleen, Lin, YouLing
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Do, Anh Hong
• Application Number:
  US08/602,998
• Time in Patent Office: 683 Days
• Field of Search: 382/276, 382/141, 382/282, 356/394
• US Class Current: 382/276
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06T 7/001 using an image reference approach
• 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 involves capturing visual indicia as a digital image, isolating the visual indicia on the digital image, converting the visual indicia to image primitives, identifying and presenting the visual indicia. An apparatus contains a camera for capturing a visual image, a video-to-digital converter, and a computer with a processor, appropriate memory, and software.

Claims

1. A method for identifying visual indicia on a surface comprising the steps of:

capturing a visual image of a scope-of-view window containing the visual indicia as a digital image;

isolating the visual indicia on the digital image;

converting the isolated visual indicia on the digital image to image primitives;

said step of converting the isolated visual indicia on the digital image to image primitives comprising the steps of;

determining an intensity trough value for the digital image;

dividing the scope-of-view window into a predetermined number of token windows;

determining the number of qualified pixels having an intensity value above said trough value in each token window; and

comparing the number of qualified pixels in each token window to a predetermined threshold value and assigning a primitive value of "1" for each token window having more qualified pixels in the token window than a predetermined threshold value and assigning "0" for each token window having less qualified pixels than the predetermined threshold value; and

comparing the image primitives to a plurality of sets of grammar primitives each corresponding to known visual indicia to find the known visual indicia that is most similar to the visual indicia of the visual image.

2. The method of claim 1, wherein the step of comparing the image primitives to a plurality of grammar primitives comprises the steps of:

comparing the image primitives to each of a plurality of sets of grammar primitives in a grammar database;

maintaining an error count of the number of differences between each comparison of grammar primitives to the image primitives; and

identifying the visual indicia as the known visual indicia having grammar primitives most similar to the image primitives.

3. The method of claim 1, further comprising producing a signal representing the known visual indicia associated with grammar primitives having the fewest differences to the image primitives for the plurality of grammar primitives.

4. The method of claim 1, further comprising the step of learning new grammar primitives by defining the image primitives as a new set of grammar primitives in response to input by an operator.

5. The method of claim 1 wherein said step of determining an intensity trough value for the digital image comprises the steps of:

dividing said scope-of-view window into a predetermined number of sampling sites;

determining the trough value for each sampling site; and

interpolating said trough value from said sampling sites.

6. A method for identifying visual indicia on a surface comprising the steps of:

capturing a visual image of a scope-of-view window containing the visual indicia as a digital image;

isolating the visual indicia on the digital image;

converting the isolated visual indicia on the digital image to image primitives;

said step of converting the isolated visual indicia on the digital image to image primitives comprises the steps of;

determining an intensity trough value for the digital image;

performing a skew adjustment;

dividing the scope-of-view window into a predetermined number of token windows;

determining the number of qualified pixels having an intensity value above said trough value in each token window; and

comparing the number of qualified pixels in each token window to a predetermined threshold value and assigning a primitive value of "1" for each token window having more qualified pixels in the token window than the predetermined threshold value and assigning "0" for each token window having less qualified pixels than the predetermined threshold value; and

comparing the image primitives to a plurality of sets of grammar primitives each corresponding to known visual indicia to find the known visual indicia that is most similar to the visual indicia of the visual image.

7. A method of identifying visual characters against a noisy background on a a surface comprising the steps of:

capturing a video image of a scope-of-view window containing the characters;

converting the video image to a digital image by using a video-to-digital converter;

isolating the visual characters on the digital image;

converting the visual characters to image primitives comprising the steps of;

determining an intensity trough value for the digital image;

dividing each visual character into a predetermined number of token windows;

determining the number of qualified pixels having an intensity value above said trough value in each token window; and

comparing the number of qualified pixels in each token window to a predetermined threshold value and assigning a primitive value of "1" for each token window having more qualified pixels in the token window than the predetermined threshold value and assigning "0" for each token window having less qualified pixels than the predetermined threshold value; and

comparing the image primitives for each character to a plurality of sets of grammar primitives in a grammar database containing known characters to determine the known character whose grammar primitives are least different from the image primitives of each character; and

identifying each visual character with the known character whose grammar primitives were least different from the image primitives.

8. The method of claim 7 including after the step of determining an intensity trough value the step of performing a skew adjustment.

9. The method of claim 7, further comprising the step of producing an output signal corresponding to the characters whose grammar primitives were least different from the image primitives.