Method, apparatus, and computer-readable storage medium for area identification

Patent metadata

Field	Value
Publication number	US-10095949-B2
Application number	US-201615298340-A
Country	US
Kind code	B2
Filing date	Oct 20, 2016
Priority date	Oct 30, 2015
Publication date	Oct 9, 2018
Grant date	Oct 9, 2018

Abstract

A method for area identification includes identifying a plurality of candidate predefined edges in a document image of a document. The candidate predefined edges are edges in a predefined direction of the document. The method further includes determining one of the candidate predefined edges to be a target predefined edge and identifying at least one information area in the document image based on the target predefined edge.

First claim

What is claimed is:

1. A method for area identification, comprising:
identifying a plurality of candidate predefined edges in a document image of a document, the candidate predefined edges being edges in a predefined direction of the document;
determining one of the candidate predefined edges to be a target predefined edge; and
identifying at least one information area in the document image based on the target predefined edge,
wherein determining one of the candidate predefined edges to be the target predefined edge includes:
sorting the candidate predefined edges to obtain sorted candidate predefined edges;
setting one of the sorted candidate predefined edges as a trial predefined edge;
performing an attempted identification of a target information area in the document image using the trial predefined edge and a relative positional relation between the target predefined edge and the target information area;
if the attempted identification using the trial predefined edge is successful:
determining the trial predefined edge to be the target predefined edge; and
if the attempted identification using the trial predefined edge is not successful:
setting a next one of the sorted candidate predefined edge as a new trial predefined edge; and
repeating the attempted identification of the target information area using the new trial predefined edge and the relative positional relation.

2. The method of claim 1 , wherein sorting the candidate predefined edges includes:
intersecting each of the candidate predefined edges with foreground color pixels at a same position of the candidate predefined edge in a processed document image, to obtain a number of intersection points corresponding to the candidate predefined edge, the processed document image being obtained by subjecting the document image to Sobel horizontal filtering and binarization; and
sorting the candidate predefined edges based on the numbers of intersection points in a descending order.

3. The method of claim 1 , wherein performing the attempted identification includes:
locating an area of interest in the document image using the trial predefined edge and the relative positional relation; and
determining whether a character area satisfying predefined characteristics exists in the area of interest, the predefined characteristics being characteristics possessed by the character area in the target information area.

4. The method of claim 3 , wherein determining whether the character area satisfying the predefined characteristics exists in the area of interest includes:
binarizing the area of interest to obtain a binarized area of interest;
calculating a horizontal histogram for the binarized area of interest in a horizontal direction of the binarized area of interest, a vertical direction in the horizontal histogram representing vertical coordinates of pixels in the binarized area of interest, and a horizontal direction in the horizontal histogram representing a number of foreground color pixels in each row of pixels having a same vertical coordinate;
calculating a vertical histogram for the binarized area of interest in a vertical direction of the binarized area of interest, a horizontal direction in the vertical histogram representing horizontal coordinates of the pixels in the binarized area of interest, and a vertical direction in the vertical histogram representing a number of foreground color pixels in each column of pixels having the same horizontal coordinate;
determining a height of a consecutive row set in the horizontal histogram and a number of consecutive column sets in the vertical histogram, the consecutive row set being a set of consecutive rows of pixels each having a number of foreground color pixels larger than a first threshold, and a consecutive column set being a set of consecutive columns of pixels each having a number of foreground color pixels larger than a second threshold; and
determining whether the height of the consecutive row set satisfies a predefined height range and whether the number of consecutive column sets satisfies a predefined number to determine whether the character area satisfying the predefined characteristics exists in the area of interest.

5. The method of claim 1 , wherein identifying the plurality of candidate predefined edges in the document image includes:
subjecting the document image to Sobel horizontal filtering and binarization to obtain a processed document image;
conducting a straight line detection in a predefined area in the processed document image to obtain a plurality of straight lines; and
identifying the straight lines as the candidate predefined edges.

6. The method of claim 1 , wherein identifying the at least one information area in the document image based on the target predefined edge includes:
determining an information area based on the target predefined edge and a relative positional relation between the target predefined edge and the information area.

7. A device for area identification, comprising:
a processor; and
a memory storing instructions that, when executed by the processor, cause the processor to:
identify a plurality of candidate predefined edges in a document image of a document, the candidate predefined edges being edges in a predefined direction of the document;
determine one of the candidate predefined edges to be a target predefined edge; and
identify at least one information area in the document image based on the target predefined edge,
wherein the instructions, when executed by the processor, further cause the processor to:
sort the candidate predefined edges to obtain sorted candidate predefined edges;
set one of the sorted candidate predefined edges as a trial predefined edge;
perform an attempted identification of a target information area in the document image using the trial predefined edge and a relative positional relation between the target predefined edge and the target information area;
if the attempted identification using the trial predefined edge is successful:
determine the trial predefined edge to be the target predefined edge; and
if the attempted identification using the trial predefined edge is not successful:
set a next one of the sorted candidate predefined edge as a new trial predefined edge; and
repeat the attempted identification of the target information area using the new trial predefined edge and the relative positional relation.

8. The device of claim 7 , wherein the instructions, when executed by the processor, further cause the processor to:
intersect each of the candidate predefined edges with foreground color pixels at a same position of the candidate predefined edge in a processed document image, to obtain a number of intersection points corresponding to the candidate predefined edge, the processed document image being obtained by subjecting the document image to Sobel horizontal filtering and binarization; and
sort the candidate predefined edges based on the numbers of intersection points in a descending order.

9. The device of claim 7 , wherein the instructions, when executed by the processor, further cause the processor to:
locate an area of interest in the document image using the trial predefined edge and the relative positional relation; and
determine whether a character area satisfying predefined characteristics exists in the area of interest, the predefined characteristics being characteristics possessed by the character area in the target information area.

10. The device of claim 9 , wherein the instructions, when executed by the processor, further cause the processor to:
binarize the area of interest to obtain a binarized area of interest;
calculate a horizontal histogram for

Assignees

• Xiaomi Inc

Inventors

• Long Fei
• Zhang Tao
• Chen Zhijun

Classifications

• G06V30/40 Primary

  Document-oriented image-based pattern recognition · CPC title

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06V10/758

  Involving statistics of pixels or of feature values, e.g. histogram matching · CPC title

• G06T2207/30176

  Document · CPC title

• G06K2209/01

  Physics · mapped topic