Automatic separation of text from background in scanned images of complex documents

• US 5,280,367 A
• Filed: 05/28/1991
• Issued: 01/18/1994
• Est. Priority Date: 05/28/1991
• Status: Expired due to Term

• Current Assignee: Hewlett-Packard Company (HP Inc.)
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Zuniga, Oscar A.
• Primary Examiner(s): Coles, Sr., Edward L.
• Assistant Examiner(s): BACARES, RAFAEL D
• Application Number:
  US07/705,838
• Time in Patent Office: 966 Days
• Field of Search: 382/51, 382/18, 382/9, 358/462, 358/464, 358/465, 358/466, 358/455, 358/456, 358/458, 358/75, 358/433, 358/475
• US Class Current: 358/462
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables
• CPC Class Codes: H04N 1/38 Circuits or arrangements for blanking or otherwise eliminating unwanted parts of pictures H04N1/387 takes precedence
• CPC Class Codes: H04N 1/40062 Discrimination between different image types, e.g. two-tone, continuous tone
• CPC Class Codes: H04N 1/4074 using histograms

Abstract

A system that converts a scanned image of a complex document into an image where text has been preserved and separated from the background. The system first subdivides the scanned image into blocks and then examines each block pixel by pixel to construct a histogram of the gray scale values of the pixels. The histogram is partitioned into a first, middle and last regions. If one or more peaks occur in the first and last regions, and a single histogram peak occurs within the middle region, the pixels are reexamined to determine the frequency of occurrence of pixels having a gray scale level of the middle peak nearby pixels which have a level of a first region peak. If this frequency is high, the middle peak is assumed to be background information. After determining the threshold, the system rescans the block applying the threshold to separate the text from background information within the block.

Claims

1. A computer implemented process for separating text information from background information in a scanned electronic image of a document, said computer implemented process comprising the steps of:

(a) electronically scanning the document to convert the document into said scanned electronic image of the document;

(b) examining said scanned electronic image and dividing said scanned electronic image into a plurality of blocks;

(c) constructing a histogram of gray scale values of pixels within one of said blocks;

(d) dividing said histogram into three regions comprising a first region, a middle region and a last region;

(d) determining a number of peaks of said histogram in each of said three regions;

(f) if said histogram contains no peak in said middle region, setting a threshold gray scale level between a gray scale level of a peak having a highest gray scale level in said first region and a gray scale level of a peak having a lowest gray scale level in said last region;

(g) separating said text information from said background information by reexamining said block using said threshold gray scale level set in step (f); and

(h) repeating steps (c) through (g) for each of said plurality of blocks.

2. The computer implemented process of claim 1 wherein step (d) further comprises the step of dividing said histogram into three equal sized regions.

3. The computer implemented process of claim 1 wherein step (f) further comprises the steps of:

(f1) if said histogram contains only a single peak in said middle region performing the steps of

(f1a) selecting a peak having a highest gray scale level in said first region,

(f1b) selecting each pixel in said block having a gray scale level equal to said gray scale level of said peak selected in step (f1a),

(f1c) for each pixel selected in step (f1b), selecting a second pixel located a predetermined distance from said selected pixel,

(f1d) if said second pixel has a gray scale level equal to a gray scale level of said single peak in said middle region, incrementing a counter,

(f1e) if said counter has a count greater than a predetermined number, setting a threshold gray scale level to a gray scale level greater than said gray scale level of said single peak in said middle region, and

(f1f) if said counter has a count less that said predetermined number, setting said threshold gray scale level to a gray scale level less than said gray scale level of said single peak in said middle region.

4. The computer implemented process of claim 1 wherein step (f) further comprises the steps of:

(f2) if said histogram contains at least two peaks in said middle region, performing the steps of

(f2a) selecting two peaks of said at least two peaks wherein said selected two peaks have a lowest and next lowest gray scale levels of said at least two peaks in said middle region, and

(f2b) setting a threshold gray scale level between said gray scale levels of said two peaks selected in step (f2a).

5. The computer implemented process of claim 1 wherein step (e) further comprises the steps of:

(e1) retrieving a histogram value for a first histogram gray scale level within a region;

(e2) retrieving a histogram value for second histogram gray scale level within said region wherein said second histogram gray scale level is one level less that said gray scale level of said first histogram gray scale level;

(e3) retrieving a histogram value for third histogram gray scale level within said region wherein said third histogram gray scale level is one level more that said gray scale level of said first histogram gray scale level;

(e4) selecting said first histogram gray scale level as a peak if a difference between twice said first histogram value and a sum of said second histogram value and said third histogram value is greater than a predetermined noise value.

6. A computer implemented process for separating text information from background information in a scanned electronic image of a document, said computer implemented process comprising the steps of:

(a) electronically scanning the document to convert the document into said scanned electronic image of the document;

(b) examining said scanned electronic image and dividing said scanned electronic image into a plurality of blocks;

(c) constructing a histogram of gray scale values of pixels within one of said blocks;

(d) dividing said histogram into three regions, comprising a first region, a middle region and a last region;

(e) locating all peaks in said histogram;

(f) removing all except a highest peak in said first region;

(g) removing all except a lowest peak in said last region;

(h) determining a number of peaks remaining in said histogram;

(i) if said histogram contains only two peaks, setting a threshold gray scale level between a gray scale level of a first peak and a gray scale level of a second peak;

(j) separating said text information from said background information by reexamining said block using said threshold gray scale level set in step (i); and

(k) repeating steps (c) through (j) for each of said plurality of blocks.

7. The computer implemented process of claim 6 wherein step (d) further comprises the step of dividing said histogram into three equal sized regions.

8. The computer implemented process of claim 6 wherein step (i) further comprises the steps of:

(i1) if said histogram contains only a single peak in said middle region performing the steps of

(i1a) selecting a peak having a highest gray scale level in said first region,

(i1b) selecting each pixel in said block having a gray scale level equal to said gray scale level of said peak selected in step (i1a),

(i1c) for each pixel selected in step (i1b), selecting a second pixel located a predetermined distance from said selected pixel,

(i1d) if said second pixel has a gray scale level equal to a gray scale level of said single peak in said middle region, incrementing a counter,

(i1e) if said counter has a count greater than a predetermined number, setting a threshold gray scale level to a gray scale level greater than said gray scale level of said single peak in said middle region, and

(i1f) if said counter has a count less that said predetermined number, setting said threshold gray scale level to a gray scale level less than said gray scale level of said single peak in said middle region.

9. The computer implemented process of claim 6 wherein step (i) further comprises the steps of:

(i2) if said histogram contains at least two peaks in said middle region, performing the steps of

(i2a) selecting two peaks of said at least two peaks wherein said selected two peaks have a lowest and next lowest gray scale levels of said at least two peaks in said middle region, and

(i2b) setting a threshold gray scale level between said gray scale levels of said two peaks selected in step (i2a).

10. The computer implemented process of claim 6 wherein step (e) further comprises the steps of:

(e1) retrieving a histogram value for a first histogram gray scale level within a region;

(e2) retrieving a histogram value for second histogram gray scale level within said region wherein said second histogram gray scale level is one level less that said gray scale level of said first histogram gray scale level;

(e3) retrieving a histogram value for third histogram gray scale level within said region wherein said third histogram gray scale level is one level more that said gray scale level of said first histogram gray scale level;

(e4) selecting said first histogram gray scale level as a peak in a difference between twice said first histogram value and a sum of said second histogram value and said third histogram value is greater than a predetermined noise value.

11. A computer implemented process for separating text information from background information in a scanned electronic image of a document, said computer implemented process comprising the steps of:

(a) electronically scanning the document to convert the document into said scanned electronic image of the document;

(b) examining said scanned electronic image and constructing a histogram of gray scale values of pixels within said scanned electronic image;

(c) dividing said histogram into three regions comprising a first region, a middle region and a last region;

(d) determining a number of peaks of said histogram in each of said three regions;

(e) if said histogram contains no peak in said middle region, setting a threshold gray scale level between a gray scale level of a peak having a highest gray scale level in said first region and a gray scale level of a peak having a lowest gray scale level in said last region;

(f) separating said text information from said background information by reexamining said scanned electronic image using said threshold gray scale level set in step (e).

12. The computer implemented process of claim 11 wherein step (c) further comprises the step of dividing said histogram into three equal sized regions.

13. The computer implemented process of claim 11 wherein step (e) further comprises the steps of:

(e1) if said histogram contains only a single peak in said middle region performing the steps of

(e1a) selecting a peak having a highest gray scale level in said first region,

(e1b) selecting each pixel in said scanned electronic image having a gray scale level equal to said gray scale level of said peak selected in step (e1a),

(e1c) for each pixel selected in step (e1b), selecting a second pixel located a predetermined distance from said selected pixel,

(e1d) if said second pixel has a gray scale level equal to a gray scale level of said single peak in said middle region, incrementing a counter,

(e1e) if said counter has a count greater than a predetermined number, setting a threshold gray scale level to a gray scale level greater than said gray scale level of said single peak in said middle region, and

(e1f) if said counter has a count less that said predetermined number setting said threshold gray scale level to a gray scale level less than said gray scale level of said single peak in said middle region.

14. The computer implemented process of claim 11 wherein step (e) further comprises the steps of:

(e2) if said histogram contains at least two peaks in said middle region, performing the steps of

(e2a) selecting two peaks of said at least two peaks wherein said selected two peaks have a lowest and next lowest gray scale levels of said at least two peaks in said middle region, and

(e2b) setting a threshold gray scale level between said gray scale levels of said two peaks selected in step (e2a).

15. The computer implemented process of claim 11 wherein step (d) further comprises the steps of:

(d1) retrieving a histogram value for a first histogram gray scale level within a region;

(d2) retrieving a histogram value for second histogram gray scale level within said region wherein said second histogram gray scale level is one level less that said gray scale level of said first histogram gray scale level;

(d3) retrieving a histogram value for third histogram gray scale level within said region wherein said third histogram gray scale level is one level more that said gray scale level of said first histogram gray scale level;

(d4) selecting said first histogram gray scale level as a peak if a difference between twice said first histogram value and a sum of said second histogram value and said third histogram value is greater than a predetermined noise value.