Method and system for dropping template from a filled in image

• US 6,507,671 B1
• Filed: 12/09/1999
• Issued: 01/14/2003
• Est. Priority Date: 12/11/1998
• Status: Expired due to Term

• Current Assignee: LinkedIn Corporation (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Zlotnick, Aviad, Kagan, Oren
• Primary Examiner(s): Wu, Jingge
• Application Number:
  US09/457,582
• Time in Patent Office: 1,132 Days
• Field of Search: 382/168-180, 382/224-239, 382/209, 382/48, 382/221, 358/450-464, 707/500, 707/530-541
• US Class Current: 382/173
• 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

In an image processing system for processing filled in forms, a method and system are disclosed for removing the fixed part or template of a form without altering the variable part or information filled into the form. Whenever the variable part overlaps a fixed part of a filled in form a straightforward removal of the template creates gaps in the information characters making them unrecognizable by Optical Character Recognition (OCR) Software. This problem is overcome, according to the invention, by scanning a filled in form using any appropriate optical device capable of capturing images made of gray level pixels. First the fixed template pixels and the variable information pixels are separated from each other based on pixel position. Thereafter, the fixed and variable part populations of pixels are further analyzed so that gaps can be filled by moving pixels from the fixed part to the variable part on the basis of their gray levels thus, permitting to filling the gaps.

Claims

1. A method for dropping the fixed part of an image of a filled in form document without altering the variable part of the image, said method comprising the steps of:

scanning the filled in form document with a device capable of generating an image of the form comprising picture elements, each picture element defining a level of gray representing a particular brightness of said picture element;

storing the image of the filled in form in gray levels;

retrieving a previously stored image of a fixed part of an image of an unfilled form document;

registering the image of the filled in form over the image of the fixed part for distinguishing between the variable part and the fixed part picture elements from their position;

collecting the picture elements positioned within the fixed part;

a. excluding those picture elements near the variable part;

b. morphologically thinning the fixed part by excluding those of the picture elements positioned at the periphery;

and establishing a first statistic of fixed part picture element levels of gray;

collecting the remaining picture elements, all belonging to the variable part, and establishing a second statistic of variable part levels of gray;

comparing said first and second statistics;

moving those of the fixed part picture elements, statistically significantly different, to the variable part.

2. A method of claim 1, where comparison of said first and second statistics and the moving of fixed part picture elements includes the steps of:

finding a first gray level threshold statistically separating the variable part picture elements from the fixed part picture elements;

moving all picture elements belonging to the fixed part population, darker than the threshold, to the variable part population.

3. The method of claim 2, further including the steps of:

preprocessing the gray level image, selecting groups of neighbor picture elements, weighting their gray level values to a common value;

collecting the picture elements positioned near the variable part, within the fixed part, establishing a third statistic on their levels of gray;

collecting all the picture elements within the updated population of the variable part, establishing a fourth statistic on their levels of gray;

finding a second gray level threshold, computed from the third and fourth statistics, statistically separating the variable part picture elements from the fixed part picture elements;

moving all picture elements belonging to the fixed part population, darker than the second threshold and touching variable part picture elements, to the variable part population.

4. The method of claim 2 further including the steps of:

counting the total number n_fixed of picture elements within the fixed part population;

counting the total number n_filled of picture elements within the variable part population;

counting the number of picture elements fixed[t] less than or equal to a given level of gray [t] in the fixed part population;

counting the number of picture elements filled[t] less than or equal to a given level of gray [t] in the variable part population;

computing the probability p[t] for a filled part picture element having a gray level less than or equal to a given value

5. The method of claim 4 further including the step of qualifying the threshold by checking that the number of picture elements below said threshold, in the fixed part population, is less than half the total number of picture elements so that following holds:

6. The method of claim 1 further including the steps of:

grouping picture elements in squares of adjacent picture elements;

finding the darkest picture element within the group;

weighting the gray level of all picture elements within the group to the one of the darkest picture elements.

7. The method of claim 1 further comprising the steps of:

splitting the whole image into a plurality of subportions;

processing independently each subportion according to the steps of claim 1;

reassembling the subportions to create the variable part.

8. The method of claim 1 further comprising:

after processing of the variable part has been completed, generating said variable part in binary levels for the purpose of being permanently stored.

9. A system for dropping the fixed part of an image of a filled in form document without altering the variable part of the image, said method comprising:

means for scanning the filled in form document with a device capable of generating an image of the form comprising picture elements, each picture element defining a level of gray representing a particular brightness of said picture element;

means for storing the image of the filled in form in gray levels;

means for retrieving a previously stored image of a fixed part of an image of an unfilled form document;

means for registering the image of the filled in form over the image of the fixed part for distinguishing between the variable part and the fixed part picture elements from their position;

means for collecting the picture elements positioned within the fixed part;

means for excluding those picture elements within the fixed part that are near the variable part;

means for morphologically thinning the fixed part by excluding those of the picture elements positioned at the periphery; and

means for establishing a first statistic of fixed part picture element levels of gray;

means for collecting the remaining picture elements, all belonging to the variable part, and establishing a second statistic of variable part levels of gray;

means for comparing said first and second statistics;

means for moving those of the fixed part picture elements, statistically significantly different, to the variable part.

10. A system of claim 9, where the means for comparing said first and second statistics comprises:

means for finding a first gray level threshold statistically separating the variable part picture elements from the fixed part picture elements;

and where the means for moving of fixed part picture elements comprises;

means for moving all picture elements belonging to the fixed part population, darker than the threshold, to the variable part population.

11. The system of claim 10, further comprising:

means for preprocessing the gray level image;

means for selecting groups of neighbor picture elements;

means for weighting the gray level values of a group of neighbor picture elements to a common value;

means for collecting the picture elements positioned near the variable part, within the fixed part;

means for establishing a third statistic on their levels of gray;

means for collecting all the picture elements within the updated population of the variable part;

means for establishing a fourth statistic on the levels of gray of the updated population of the variable part;

means for finding a second gray level threshold, computed from the third and fourth statistics, statistically separating the variable part picture elements from the fixed part picture elements;

means for moving all picture elements belonging to the fixed part population, darker than the second threshold and touching variable part picture elements, to the variable part population.

12. The system of claim 10 further comprising:

means for counting the total number n_fixed of picture elements within the fixed part population;

means for counting the total number n_filled of picture elements within the variable part population;

means for counting the number of picture elements fixed[t] less than or equal to a given level of gray [t] in the fixed part population;

means for counting the number of picture elements filled[t] less than or equal to a given level of gray [t] in the variable part population;

means for computing the probability p[t] for a filled part picture element having a gray level less than or equal to a given value

13. The system of claim 12 further comprising:

means for qualifying the threshold by checking that the number of picture elements below said threshold, in the fixed part population, is less than half the total number of picture elements so that following holds;

14. The system of claim 9 further comprising:

means for grouping picture elements in squares of adjacent picture elements;

means for finding the darkest picture element within the group;

means for weighting the gray level of all picture elements within the group to the one of the darkest picture elements.

15. The system of claim 9 further comprising:

means for splitting the whole image into a plurality of subportions;

means for processing independently each subportion using the means of claim 9; and

means for reassembling the subportions to create the variable part.

16. The system of claim 9 further comprising:

means for generating said variable part in binary levels for the purpose of being permanently stored after processing of the variable part has been completed.