Method, device, and computer-readable storage medium for area extraction

Patent metadata

Field	Value
Publication number	US-10127471-B2
Application number	US-201615298352-A
Country	US
Kind code	B2
Filing date	Oct 20, 2016
Priority date	Oct 30, 2015
Publication date	Nov 13, 2018
Grant date	Nov 13, 2018

Abstract

A method for area extraction includes obtaining an area position of a first information area in a document image, determining a second information area according to the area position of the first information area, and segmenting the second information area to obtain at least one character zone.

First claim

What is claimed is:

1. A method for area extraction, comprising:
obtaining an area position of a first information area in a document image;
determining a second information area according to the area position of the first information area; and
segmenting the second information area to obtain at least one character zone,
wherein segmenting the second information area comprises:
performing a binarization processing on the second information area to obtain a binarized second information area;
calculating a horizontal histogram of the binarized second information area in a horizontal direction, a vertical direction in the horizontal histogram representing vertical coordinates of pixels in the binarized second information area, and a horizontal direction in the horizontal histogram representing a number of foreground color pixels in each row of pixels;
recognizing one or more text rows based on consecutive row sets in the horizontal histogram, each of the consecutive row sets including consecutive rows each having the number of foreground color pixels larger than a first threshold;
calculating a vertical histogram for a target text row in the one or more text rows in a vertical direction, a horizontal direction in the vertical histogram representing horizontal coordinates of the pixels in the target text row, and a vertical direction in the vertical histogram representing the number of foreground color pixels in each column of pixels; and
recognizing the at least one character zone based on consecutive column sets in the vertical histogram, each of the consecutive column sets including consecutive columns each having the number of foreground color pixels larger than a second threshold, and

wherein the method further comprises:
recognizing a row spacing between two adjacent text rows based on the consecutive row sets in the horizontal histogram; and
discarding, if the row spacing is larger than a third threshold, one of the two adjacent text rows that is closer to an upper edge or a lower edge of the second information area.

2. The method of claim 1 , wherein:
obtaining the area position includes:
obtaining two vertex coordinate pairs representing the area position, and determining the second information area includes:
determining the second information area according to the two vertex coordinate pairs and a predetermined relative position relationship between the two vertex coordinate pairs and the second information area.

3. The method of claim 2 , wherein determining the second information area according to the two vertex coordinate pairs and the predetermined relative position relationship includes:
determining a lower edge of the second information area according to a vertical coordinate in a closer vertex coordinate pair of the two vertex coordinate pairs that is closer to the second information area;
determining an upper edge of the second information area according to the vertical coordinate in the closer vertex coordinate pair and a predetermined height;
determining a left edge of the second information area according to a horizontal coordinate of any one of the two vertex coordinate pairs and a first predetermined width;
determining a right edge of the second information area according to a horizontal coordinate of any one of the two vertex coordinate pairs and a second predetermined width; and
determining the second information area according to the lower edge, the upper edge, the left edge, and the right edge.

4. The method of claim 1 , further comprising:
recognizing a character spacing between two adjacent character zones based on the consecutive column sets in the vertical histogram;
recognizing, if the two adjacent character zones are on the left of the second information area and the character spacing is larger than a fourth threshold, a right one of the two adjacent character zones as a first character zone in the target text row; and
recognizing, if the two adjacent character zones are on the right of the second information area and the character spacing is larger than a fifth threshold, a left one of the two adjacent character zones as a last character zone in the target text row.

5. A device for area extraction, comprising:
a processor; and
a memory storing instructions that, when executed by the processor, cause the processor to:
obtain an area position of a first information area in a document image;
determine a second information area according to the area position of the first information area; and
segment the second information area to obtain at least one character zone,

wherein in segmenting the second information area to obtain at least one character zone, the instructions further cause the processor to:
perform a binarization processing on the second information area to obtain a binarized second information area;
calculate a horizontal histogram of the binarized second information area in a horizontal direction, a vertical direction in the horizontal histogram representing vertical coordinates of pixels in the binarized second information area, and a horizontal direction in the horizontal histogram representing a number of foreground color pixels in each row of pixels;
recognize one or more text rows based on consecutive row sets in the horizontal histogram, each of the consecutive row sets including consecutive rows each having the number of foreground color pixels larger than a first threshold;
calculate a vertical histogram for a target text row in the one or more text rows in a vertical direction, a horizontal direction in the vertical histogram representing horizontal coordinates of the pixels in the target text row, and a vertical direction in the vertical histogram representing the number of foreground color pixels in each column of pixels; and
recognize the at least one character zone based on consecutive column sets in the vertical histogram, each of the consecutive column sets including consecutive columns each having the number of foreground color pixels larger than a second threshold, and

wherein the instructions further cause the processor to:
recognize a row spacing between two adjacent text rows based on the consecutive row sets in the horizontal histogram; and
discard, if the row spacing is larger than a third threshold, one of the two adjacent text rows that is closer to an upper edge or a lower edge of the second information area.

6. The device of claim 5 , wherein the instructions further cause the processor to:
obtain two vertex coordinate pairs representing the area position; and
determine the second information area according to the two vertex coordinate pairs and a predetermined relative position relationship between the two vertex coordinate pairs and the second information area.

7. The device of claim 6 , wherein the instructions further cause the processor to:
determine a lower edge of the second information area according to a vertical coordinate in a closer vertex coordinate pair of the two vertex coordinate pairs that is closer to the second information area;
determine an upper edge of the second information area according to the vertical coordinate in the closer vertex coordinate pair and a predetermined height;
determine a left edge of the second information area according to a horizontal coordinate of any one of the two vertex coordinate pairs and a first predetermined width;
determine a right edge of the second information area according to a horizontal coordinate of any one of the two vertex coordinate pairs and a second predetermined width; and
determine the second information area according to the lower edge, the upper edge, the left edge, and the right edge.

8. The device of claim 5 , wherein the instructions further cause the proce

Assignees

• Xiaomi Inc

Inventors

• Long Fei
• Zhang Tao
• Chen Zhijun

Classifications

• G06V30/412 Primary

  Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables · CPC title

• G06V10/26

  Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion · CPC title

• G06V10/22

  by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition · CPC title

• G06V10/507

  Summing image-intensity values; Histogram projection analysis · CPC title

• G06T7/11

  Region-based segmentation · CPC title