Automated portrait/landscape mode detection on a binary image

• US 5,592,572 A
• Filed: 11/05/1993
• Issued: 01/07/1997
• Est. Priority Date: 11/05/1993
• Status: Expired due to Term

• Current Assignee: United States Department Of Health And Human Services (Government of the United States of America)
• Original Assignee: United States Department Of Health And Human Services (Government of the United States of America)
• Inventors: Le, Daniel X. D.
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Bella, Matthew C.
• Application Number:
  US08/147,798
• Time in Patent Office: 1,159 Days
• Field of Search: 382/9, 382/18, 382/46, 382/51, 382/28, 382/289, 382/298, 382/171, 382/173, 382/176, 382/194
• US Class Current: 382/289
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/1478 of characters or characters lines
• CPC Class Codes: H04N 1/00795 Reading arrangements details of scanning heads H04N1/024; scanning arrangements therefor H04N1/04

Abstract

A process for determining an orientation of a binary image includes segmentation of the image into many square regions. Orientations of the individual regions that are determined to be textual squares allow local variations of the binary image to establish an overall orientation for the binary image. The process iteratively groups and consolidates the individual regions into successively larger and larger regions. An orientation for particular ones of the larger regions is determined by a single mode of the various composite modes of the particular one regions having the greatest weight. After all the regions have been consolidated in this fashion, the process uses the orientation of the single consolidated region as the orientation of the binary image.

Claims

1. A method for determining an orientation of a binary image, wherein the binary image is generated from a scan of a document, the method comprising the steps of:

a) scanning the document to generate the binary image;

b) segmenting the binary image into a first plurality of regions, wherein bounds of regions in the first plurality of regions are independent of content of the binary images;

c) determining, for regions in the first plurality of regions, which of such regions are textual regions;

d) determining an orientation of at least each textual region of said first plurality of regions; and

e) grouping regions of said first plurality of regions into regions of a second plurality of regions;

f) determining, for regions in the second plurality of regions, which of such regions are textual regions;

g) determining an orientation of at least each textual region of the second plurality of regions; and

h) determining an orientation of the binary image from the orientations determined for the second plurality of regions.

2. The method of claim 1 wherein said step of segmenting the binary image into a first plurality of regions is a step of segmenting the binary image into uniformly-sized, square regions.

3. The method of claim 1 wherein steps d) and g) use projection histograms to determine orientations.

9. The method of claim 1, further comprising the steps of repeating steps e) though h) to consolidate the second plurality of regions into a third plurality of regions, and so on, until regions are consolidated into a single region covering the binary image and an orientation of the single region is determined, the orientation of the single region being the orientation of the binary image.

10. The method of claim 3 wherein said step of determining comprises the steps of:

assigning a combined weight to each set of a plurality of sets of neighboring regions in similarly oriented regions; and

selecting a particular orientation having a largest combined weight as an orientation for the set.

4. A method for processing a binary image to detect a page orientation, comprising the steps of:

a) scanning a document to generate the binary image;

b) segmenting the binary image into a plurality uniformly-sized of regions;

c) mapping each region of said plurality of regions into a projection histogram for that region;

d) calculating a total number of black pixels for each region of said plurality of regions;

e) categorizing each region of said plurality of regions as a textual region or a non-textual region;

f) recognizing an orientation for each textual region of said plurality of regions based on the at least one projection histogram for said each textual region;

g) consolidating regions of said plurality of regions into a first plurality of consolidated regions;

h) identifying, for each consolidated region of the first plurality of consolidated regions, a mode for each possible orientation of image portions within said each consolidated region;

i) evaluating orientation for each of said consolidated regions based on similarly oriented regions of each particular consolidated region;

j) further consolidating consolidated regions into fewer and larger consolidated regions, while identifying for said fewer and larger consolidated regions an orientation based on the orientation of regions consolidated; and

k) when a single consolidated region remains which covers the binary image, determining a page orientation for the binary image by finding an orientation of the remaining single consolidated region from an orientation associated with the orientation of regions within the single consolidated region which have the greatest weighted total mode for such orientation.

5. A method for determining an orientation of a binary image of a document, comprising the steps of:

a) scanning a document to form the binary image;

b) segmenting the binary image into a first plurality of regions which are uniformly sized;

c) determining, for each region of said first plurality of regions, whether said region is a textual region;

d) determining an orientation for each said textual region;

e) grouping a plurality of sets of neighboring regions of said first plurality of regions into a second plurality of consolidated regions which are uniformly-sized and larger than regions of the first plurality of regions;

f) determining an orientation for each particular consolidated region of said second plurality of consolidated regions using said orientations of said textual regions in the particular set of neighboring regions of said first plurality of regions making up said particular consolidated region of said second plurality of regions;

g) grouping a plurality of sets of neighboring previously consolidated regions of a plurality of consolidated regions into a lesser plurality of larger consolidated uniform regions;

h) determining an orientation for each particular consolidated region of said lesser plurality of consolidated regions using said orientations of said previously consolidated regions making up said particular consolidated region of said lesser plurality of consolidated regions;

i) repeating steps g) and h) until a single consolidated region remains; and

j) assigning said orientation of said single consolidated region as the orientation of the binary image.

6. A method for determining an orientation of a binary image of a document, comprising the steps of:

a) scanning a document to form a binary image;

b) segmenting the binary image into a jⁿ ×

kⁿ array of uniformly-sized regions, where j, k, and n are all natural numbers;

c) determining, for each region of said jⁿ ×

kⁿ array, whether said region is a textual region;

d) determining an orientation for each said textual region;

e) grouping a plurality of sets of j×

k regions of said jⁿ ×

kⁿ array into a j^n-1 ×

k^n-1 array of uniformly-sized consolidated regions;

f) determining an orientation for each consolidated region of said j^n-1 ×

k^n-1 array of uniformly-sized consolidated regions using said orientations said textual regions in said set of j×

k regions making up each said consolidated region;

g) grouping a plurality of sets of j×

k regions of said j^n-(a-1) ×

k^n-(a-1) array into a j^n-a ×

k^n-a array of uniformly-sized consolidated regions, where a is a natural number initially equal to 2;

h) determining an orientation for each consolidated region of said j^n-a ×

k^n-a array using said orientations of said previously consolidated j×

k regions making up each said consolidated region of said j^n-a ×

k^n-a array;

i) iterating steps f) and g) until a single consolidated region remains wherein a is incremented by 1 for each iteration until a equals n-1; and

j) assigning said orientation of said single consolidated region as the orientation of the binary image.

7. The orientation determining method of claim 6 further comprising a step of appending a plurality of blank regions to the binary image so as to produce a square array with j equal to k.

8. The orientation determining method of claim 7 wherein j equals 3.