System for reading data glyphs

• US 6,078,698 A
• Filed: 09/20/1999
• Issued: 06/20/2000
• Est. Priority Date: 09/20/1999
• Status: Expired due to Term

• Current Assignee: FLIR Systems, Inc. (Teledyne Technologies Incorporated)
• Original Assignee: FLIR Systems, Inc. (Teledyne Technologies Incorporated)
• Inventors: Griner, James C., Williams, Richard P., Rystrom, Larry, Jones, Creed F. III, Lorton, Christopher W., Orrell, James D. III
• Primary Examiner(s): Pitts, Harold I.
• Application Number:
  US09/399,638
• Time in Patent Office: 274 Days
• Field of Search: 235/494, 382/278
• US Class Current: 382/278
• CPC Class Codes: G06K 7/14 using light without selection of wavelength, e.g. sensing reflected white light G06K7/10831 - G06K7/1097 take precedence
• CPC Class Codes: G06K 7/143 Glyph-codes

Abstract

A high speed system for locating and decoding glyphs on documents is disclosed. The system includes acquiring one or more images of a document containing a glyph. One-dimensional projections of the images are correlated against a reference function to locate the glyph in the images. The position of the glyph is refined by correlating against a kernel designed to have a maximum response when aligned over a corner of the glyph. Symbols in the glyph are decoded utilizing a kernel which generates a positive response for one symbol type and a negative response for the other.

Claims

1. A method of locating a glyph on a document, where the glyph contains symbols positioned in a grid of rows and columns, the method comprising:

scanning the document to acquire at least one image of at least part of the document;

preparing a one-dimensional kernel function having a strong correlation response when aligned with a one-dimensional projection of a glyph;

computing a one-dimensional projection over at least part of the image, where the at least part of the at least one image includes at least part of the glyph;

correlating the one-dimensional projection with the kernel function, where the kernel is located at a selected position in the projection;

repeating the step of correlating for a plurality of relative positions of the kernel in the one dimensional projection; and

analyzing the results of the repeated correlations to locate a position where there is an extremum in correlation response.

2. The method of claim 1, wherein the kernel is a saw-tooth kernel.

3. The method of claim 1, wherein the glyph contains upstroke and downstroke symbols.

4. The method of claim 1, wherein the step of scanning includes acquiring images of plural overlapping areas of the document.

5. The method of claim 4, wherein the areas have a minimum dimension at least twice as large as the maximum dimension of the glyph.

6. The method of claim 4, wherein the areas are chosen as bands extending across one dimension of the document, and where the bands are half-overlapping with each other.

7. The method of claim 1, wherein the glyph includes symbols of a first type and a second type, each type having a predetermined pattern with the two patterns having a region of intersection, each symbol defining a cell, and further comprising the steps of:

forming a second kernel function including a plurality of cells corresponding to the symbol cells, at least a portion of the kernel cells including non-zero entries at locations corresponding to the region of intersection of the symbol cells; and

conducting a correlation of the second kernel function with a selected portion of the image for a plurality of selected portions of the image to search for a maximum correlation response.

8. The method of claim 7 further comprising the steps of forming a third kernel function, where the third kernel function includes positive entries in locations corresponding to the pattern of the first symbol type and negative entries in locations corresponding to the pattern of the second symbol type;

computing a correlation between the kernel function and a symbol cell in the glyph; and

declaring the symbol to be of the first type if the result of the correlation exceeds a first predetermined value and declaring the symbol to be of the second type if the result of the correlation is less than a second predetermined value.

9. A method of decoding symbols from an image of a glyph, where there are first and second symbol types in the glyph, each symbol type having a predetermined pattern, the method comprising:

forming a kernel function, where the kernel function includes positive entries in locations corresponding to the pattern of the first symbol type and negative entries in locations corresponding to the pattern of the second symbol type;

computing a correlation between the kernel function and a symbol in the glyph; and

declaring the symbol to be of the first type if the result of the correlation exceeds a first predetermined value and declaring the symbol to be of the second type if the result of the correlation is less than a second predetermined value.

10. The method of claim 9, wherein the first and second predetermined values are both equal to zero.

11. The method of claim 9, wherein the entries in the kernel function which do not correspond to the pattern of either symbol type are zero.

12. The method of claim 9, wherein the positive entries are all equal to one.

13. The method of claim 9, wherein the first and second symbol types are upstrokes and downstrokes, respectively.

14. The method of claim 13, wherein the kernel function has non-zero entries along diagonals thereof.

15. A method of finding the location of symbols within an image incorporating a glyph with a plurality of symbols disposed in a regular array of cells therein, where the symbols include a first type and a second type, each type having a predetermined pattern with the two patterns having a region of intersection, the method comprising:

forming a correlation kernel including a plurality of cells corresponding to the symbol cells, at least a portion of the kernel cell including non-zero entries at locations corresponding to the region of intersection of the symbol cells; and

conducting a correlation of the kernel with a selected portion of the image for a plurality of selected portions of the image to search for a maximum correlation response.

16. The method of claim 15, wherein only the locations in the correlation kernel corresponding to regions of intersection are non-zero.

17. The method of claim 15, wherein there are between 4 and 10 cells in the correlation kernel.

18. The method of claim 15, wherein the entries in cells along a first side of the correlation kernel are opposite in sign to the entries in cells not along a side of the correlation kernel.

19. The method of claim 18, wherein the entries in cells along a second side adjacent the first side of the correlation kernel are opposite in sign to the entries in cells not along a side of the correlation kernel.

20. The method of claim 15, wherein no more than one entry in each cell is non-zero.

21. The method of claim 15, wherein the non-zero entries have a magnitude of one.