Granted Patent B2

US 8,605,982 · App. 13/422,463 · Granted Dec 10, 2013

Abstract

Claims (47)

1. A method of resolving a check image region within a document image comprising the steps of:

processing by optical character recognition a full page document image containing rows of text;

defining a lower boundary of the check image region by the location of the magnetic ink character recognition data;

establishing an upper boundary of the check image region, the upper boundary defined by a collection of string literals derived from a regressive analysis of string literals spatially associated with a plurality of previously validated upper boundaries; and

digitally cropping the full page document image above the upper boundary and cropping the full page document image below the lower boundary to transform the full page document image to a check-specific image.

2. The method of claim 1 further comprising the step of creating a new image file of the check-specific image.

3. The method of claim 1 wherein the regressive analysis further comprises the steps of:

identifying the upper boundary for a processed check, the upper boundary defined by a measurement selected from the group consisting of distance below the full page document image top, distance above the full page document image bottom, and distance above the magnetic ink character recognition data;

identifying string literals spatially proximate to the upper boundary of the processed check wherein the string literals may be located above or below the upper boundary;

saving the string literal, the upper boundary measurement and a relative position of the string literal to the upper boundary to a data store;

establishing a confidence threshold for defining an otherwise unknown upper boundary based on detection of the string literal; and

defining the upper boundary of subsequently processed checks having string literals matching those saved to the data store and meeting the confidence threshold.

4. The method of claim 3 wherein the confidence threshold is established based on the frequency of the string literal successfully identifying a validated check upper boundary.

5. The method of claim 1 whereby the tallest check height dimension is selected from the collection to define the upper boundary responsive to the absence of detectable horizontal raster lines.

6. A method of resolving a check image region within a document image comprising the steps of:

processing by optical character recognition a full page document image containing rows of text;

defining a lower boundary of the check image region by the location of the magnetic ink character recognition data;

establishing a check top candidate area, the check top candidate area defined by a collection of string literals derived from a regressive analysis of string literals associated with a previously validated check area upper boundary;

detecting a horizontal raster line within the check top candidate area whereby the horizontal raster line defines the top of the check;

defining an upper boundary of the check image region by the location of the horizontal line; and

digitally cropping the full page document image above the upper boundary and cropping the full page document image below the lower boundary to transform the full page document image to a check-specific image.

7. The method of claim 6 further comprising the step of creating a new image file of the check-specific image.

8. The method of claim 6 wherein the step of defining the upper boundary of the check image region comprises locating a horizontal raster line farthest from the lower boundary from a plurality of horizontal raster lines within the check top candidate area.

9. The method of claim 6 whereby the tallest check height dimension is selected from the collection to define the upper boundary responsive to the absence of detectable horizontal raster lines.

10. The method of claim 6 further comprising the step of creating a new image file of the check-specific image.

11. The method of claim 6 wherein the regressive analysis further comprises the steps of:

identifying the upper boundary for a processed check, the upper boundary defined by a measurement selected from the group consisting of distance below the full page document image top, distance above the full page document image bottom, and distance above the magnetic ink character recognition data;

identifying string literals spatially proximate to the upper boundary of the processed check wherein the string literals may be located above or below the upper boundary;

saving the string literal, the upper boundary measurement and a relative position of the string literal to the upper boundary to a data store;

establishing a confidence threshold for defining an otherwise unknown upper boundary based on detection of the string literal; and

defining the upper boundary of subsequently processed checks having string literals matching those saved to the data store and meeting the confidence threshold.

12. One or more non-transitory, computer-readable media having computer-executable instructions for performing a method of running a software program on a computer, the computer operating under an operating system, the method including issuing instructions from the software program comprising:

processing by optical character recognition a full page document image containing rows of text;

defining a lower boundary of the check image region by the location of the magnetic ink character recognition data;

establishing an upper boundary of the check image region, the upper boundary defined by a collection of string literals derived from a regressive analysis of string literals spatially associated with a previously validated upper boundary; and

digitally cropping the full page document image above the upper boundary and cropping the full page document image below the lower boundary to transform the full page document image to a check-specific image.

13. The one or more non-transitory, computer-readable media of claim 12 further comprising the step of creating a new image file of the check-specific image.

14. The one or more non-transitory, computer-readable media of claim 12 wherein the step of defining the upper boundary of the check image region comprises locating a horizontal raster line farthest from the lower boundary from a plurality of horizontal raster lines within a check top candidate area defined by an identified string literal.

15. The one or more non-transitory, computer-readable media of claim 12 wherein the tallest check height dimension is selected from the collection to define the upper boundary responsive to the absence of detectable horizontal raster lines or string literals.

16. The one or more non-transitory, computer-readable media of claim 12 wherein the regressive analysis comprises the steps of:

identifying the upper boundary for a processed check, the upper boundary defined by a measurement selected from the group consisting of distance below the full page document image top, distance above the full page document image bottom, and distance above the magnetic ink character recognition data;

identifying string literals spatially proximate to the upper boundary of the processed check wherein the string literals may be located above or below the upper boundary;

saving the string literal, the upper boundary measurement and a relative position of the string literal to the upper boundary to a data store;

establishing a confidence threshold for establishing an otherwise unknown upper boundary based on detection of the string literal; and

establishing the upper boundary of subsequently processed checks having string literals matching those saved to the data store and meeting the confidence threshold.

17. The method of claim 16 wherein the confidence threshold is established based on the frequency of the string literal successfully identifying a validated check upper boundary.

18. The method of claim 12 whereby the tallest check height dimension is selected from the collection to define the upper boundary responsive to the absence of detectable horizontal raster lines.

Assignments (11)