Method and apparatus for separating static and dynamic portions of document images

• US 5,631,984 A
• Filed: 04/07/1995
• Issued: 05/20/1997
• Est. Priority Date: 12/09/1993
• Status: Expired due to Term

• Current Assignee: NCR Corporation (NCR Voyix Corporation)
• Original Assignee: NCR Corporation (NCR Voyix Corporation)
• Inventors: Graf, Hans P., Mayer, Daniel J.
• Primary Examiner(s): Couso, Jose L.
• Application Number:
  US08/418,917
• Time in Patent Office: 774 Days
• Field of Search: 382/135, 382/173, 382/191, 382/276, 382/318, 382/209, 382/282, 382/283, 382/284, 235/379, 235/462
• US Class Current: 382/317
• CPC Class Codes: G06V 30/412 Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables
• CPC Class Codes: H04N 1/4115 involving the recognition of specific patterns, e.g. by symbol matching
• CPC Class Codes: H04N 1/4177 encoding document change data, e.g. form drop out data

Abstract

The present invention provides a method and apparatus for compressing images of financial instruments and other documents. The method of the present invention includes the steps of scanning a plurality of documents to obtain an electronic image of each document; identifying a static portion in the electronic image of each of the documents, containing information which remains substantially unchanged for the plurality of documents, by locating and reading a document identifier in the image; storing the document identifier in a database; identifying a dynamic portion, typically containing distinct information for each of the documents, in each of the electronic images; isolating the dynamic portion from the static portion within the image to obtain a dynamic image containing only the dynamic portion; and storing the dynamic image in the database. The present invention provides efficient techniques for identifying and isolating dynamic information in a document, such as handwritten text on a check.

Claims

1. A method for use in a document image processing system, said system processing digital signals representing electronic images of documents of a type that include both a static form and an added dynamic information, the method comprising the steps of:

receiving a first signal representing an electronic image of a substantially complete document of said type, said first signal including image data for both said static form and said added dynamic information;

automatically identifying a portion of the first signal which corresponds to a portion of the added dynamic information by electronically comparing the first signal to at least one predetermined image feature inherently characteristic of the added dynamic information;

processing the automatically identified portion of the first signal to determine an additional feature of the added dynamic information;

electronically extracting portions of said first signal having the additional feature;

generating a second signal that includes at least a subset of the extracted portions of the first signal; and

storing the second signal such that said first signal representing said image can be substantially reconstructed from said second signal and an image of the form.

2. The method of claim 1 wherein the contrast of the first signal is enhanced prior the step of identifying a portion of the first signal.

3. The method of claim 2 wherein the contrast of the first signal is enhanced by performing a histogram equalization on the first signal.

4. The method of claim 2 wherein the contrast of the first signal is enhanced by performing an edge enhancement on the image pixels in the first signal.

5. The method of claim 1 wherein the first signal represents a color image and the method further includes the step of generating a first signal representing an electronic binary image by measuring the first signal against a predetermined threshold.

6. The method of claim 1 wherein said step of identifying a portion of the first signal which corresponds to a portion of the added information further includes the steps of:

identifying portions of said first signal which correspond to strokes and edges in said added information by comparing the first signal to predetermined image features corresponding to said strokes and edges;

measuring the distance between edges in the identified portions to thereby obtain a characteristic of said strokes; and

identifying portions of the first signal which correspond to groups of strokes having said characteristic.

7. The method of claim 6 wherein the step of measuring the distance between edges in the identified portions to thereby obtain a characteristic of said strokes further includes performing an autocorrelation of feature maps indicative of the identified portions along a cross-section of the identified portions.

8. The method of claim 1 wherein said step of identifying a portion of the first signal which corresponds to a portion of the added information further includes convolving the first signal with a set of convolution kernels in a neural network.

9. The method of claim 1 wherein said step of processing the identified portion of the first signal to determine an additional feature of the added information further includes the step of determining a color spectrum range of said identified portion.

10. The method of claim 9 wherein said step of extracting portions of said first signal having the additional feature further includes filtering out components of said first signal falling outside said color spectrum range.

11. The method of claim 9 wherein the step of determining a color spectrum range further includes the steps of:

determining a plurality of color spectrum ranges from corresponding different color versions of the first signal; and

filtering out components of said first signal falling outside a selected one of said plurality of color spectrum ranges.

12. The method of claim 1 wherein said step of extracting portions of said first signal having the additional feature further includes performing connected component analysis on portions of the first signal having the additional feature.

13. The method of claim 1 wherein said step of extracting portions of said first signal having the additional feature further includes the step of electronically comparing the first signal to arrangements of image features characteristic of the additional feature as determined in the processing step.

14. The method of claim 1 wherein the step of identifying a portion of the first signal which corresponds to a portion of the added information is performed only on predetermined portions of the first signal which are most likely to include the added information.

15. An apparatus for use in a document image processing system, said system processing electronic images of documents of a type that include both a static form and an added dynamic information, the apparatus comprising;

means for receiving a first electronic image of a substantially complete document of said type, said first electronic image including image data for both said static form and said added dynamic information;

means for identifying a portion of the first electronic image which corresponds to a portion of the added information by comparing the first electronic image to at least one predetermined image feature inherently characteristic of the added dynamic information;

means for processing the identified portion of the first electronic image to determine an additional feature of the added dynamic information;

means for electronically extracting portions of said first image having the additional feature; and

means for generating a compressed image that includes at least a subset of the extracted portions of the first signal such that said first image can be substantially reconstructed from said compressed image and an electronic image of the form.

16. The apparatus of claim 15 wherein said document comprises a preprinted check form with said added information entered thereon.

17. The apparatus of claim 15 further including a digital memory in which said form image is stored.

18. The apparatus of claim 15 further including means for recombining said compressed image with said form image in order to substantially reconstruct said first image of said document.

19. The apparatus of claim 15 wherein said apparatus further includes means for generating a color image corresponding to said first image of said document.

20. The apparatus of claim 15 further comprising means for performing a histogram equalization on said first electronic image to enhance the contrast of said image.

21. The apparatus of claim 15 wherein the first electronic image is a color image and the apparatus further includes means for comparing the color image to a predetermined threshold to generate a binary first image therefrom.

22. The apparatus of claim 15 wherein the means for identifying a portion of the first electronic image which corresponds to a portion of the added information further includes:

means for identifying portions of said first electronic image which correspond to strokes and edges in said added information by comparing the first electronic image to predetermined image features corresponding to said strokes and edges;

means for measuring the distance between edges in the identified portions to thereby obtain a characteristic of said strokes; and

means for identifying portions of the first electronic image which correspond to groups of strokes having said characteristic.

23. The apparatus of claim 15 wherein said means for identifying a portion of the first electronic image which includes a neural network which convolves the first electronic image with a set of convolution kernels.

24. The apparatus of claim 15 wherein said means for processing the identified portion of the first electronic image to determine an additional feature of the added information further includes means for determining a color spectrum range of said identified portion.

25. The apparatus of claim 24 wherein said means for extracting portions of said first electronic image having the additional feature further includes a filter adapted to filter out components of said first electronic image falling outside said color spectrum range.

26. The apparatus of claim 15 wherein said means for extracting portions of said first electronic image having the additional features further includes:

a filter for filtering the first electronic image to remove portions thereof falling outside a color spectrum range; and

a neural network for processing the filtered first electronic image.

27. A method for use in a document image processing system, said system processing digital signals representing electronic images of documents of a type that include both a static form and an added dynamic information, the method comprising the steps

receiving a first signal representing an electronic image of a substantially complete document of said type said first signal including image data for both said static form and said added dynamic information;

determining from said received first signal an indicator of said form;

identifying a portion of the first signal which corresponds to a portion of the added dynamic information by comparing the first signal to said at least one predetermined image feature inherently characteristic of the added dynamic information;

processing the identified portion of the first signal to determine an additional feature of the added dynamic information;

electronically extracting portions of said first signal having the additional feature;

generating a second signal that includes at least a representation of said form indicator and a subset of the extracted portions of the first signal; and

storing said second signal such that said first signal representing said image can be substantially reconstructed from said second signal and an image of the form corresponding to the indicator.

28. The method of claim 27 further including the steps of:

retrieving said document form indicator from a memory in which said indicator is stored;

generating a third signal representing said image of said form corresponding to said indicator; and

reconstructing the first signal representing the image of the document by combining said second signal and said third signal.

29. The method of claim 27 wherein said step of processing the identified portion of the first signal to determine an additional feature of the added information further includes the step of determining a color spectrum range of said identified portion.

30. The method of claim 29 wherein said step of extracting portions of said first signal having the additional feature further includes filtering out components of said first signal falling outside said color spectrum range.

31. A method for use in a document image processing system, said system processing digital signals representing electronic images of documents of a type that include both a static original information portion and a dynamic added information portion, the method comprising the steps of:

receiving a first signal representing an electronic image of a substantially complete document of said type, said signal including both the original information and the dynamic added information portions of said document;

extracting from said first signal substantially all of the added information portion of said document, regardless of location thereon, by automatically distinguishing features of said added information portion from features of said original information portion; and

generating a second signal that includes at least a representation of said indicator of said original information portion and said extracted added information portion of said document whereby said first signal representing said image can be substantially reconstructed from said second signal and an image of the original information portion corresponding to said indicator.