Method and system for detecting text in digital engineering drawings

Patent metadata

Field	Value
Publication number	US-10685225-B2
Application number	US-201815900092-A
Country	US
Kind code	B2
Filing date	Feb 20, 2018
Priority date	Dec 29, 2017
Publication date	Jun 16, 2020
Grant date	Jun 16, 2020

Abstract

The present disclosure discloses method and text detection system for detecting text in digital engineering drawings. The text detection system receives digital engineering drawing, where digital image is converted to binary image. One or more lines inclined at one or more angles are detected based on pre-defined parameters, where one or more lines are removed from binary image. The text detection system determines image contours in binary image based on connected pixels with similar characteristics and identify one or more regions in binary image, to be non-text regions based on height, width, and pixel density, of corresponding image contours and predefined minimum height, minimum width, maximum height, maximum width, and predefined pixel density, where non-text regions are removed from binary image. The text detection system determines text regions in binary image based on removal of non-text regions and processing text regions to detect text in binary image of engineering drawing.

First claim

What is claimed is:

1. A method of detecting text in digital engineering drawings, the method comprising:
receiving, by a text detection system, a digital image of an engineering drawing, wherein the digital image is converted to a binary image;
detecting, by the text detection system, one or more lines inclined at one or more angles, in the binary image, based on pre-defined parameters, wherein the one or more lines are removed from the binary image;
determining, by the text detection system, one or more image contours in the binary image based on connected pixels with similar characteristics;
identifying, by the text detection system, one or more regions in the binary image, to be non-text regions based on a height, a width, and a pixel density, of corresponding one or more image contours and a predefined minimum height, minimum width, maximum height, maximum width, and a predefined pixel density, wherein the one or more non-text regions are removed from the binary image, wherein the pixel density is a ratio of number of white pixels and product of contour width and height;
determining, by the text detection system, one or more text regions in the binary image based on the removal of the one or more non-text regions; and

processing, by the text detection system, the one or more text regions to detect text in the binary image of the engineering drawing.

2. The method as claimed in claim 1 , wherein the pre-defined parameters comprise distance resolution between two lines, angle resolution of lines passing through a point, number of votes for a line, and minimum length for the line.

3. The method as claimed in claim 1 , wherein identifying the one or more regions to be non-text region comprises detecting the height of the one or more image contour corresponding to the one or more regions to be one of, greater than a predefined maximum height and lesser than a predefined minimum height, the width of the one or more image contours to be one of greater than a predefined maximum width and lesser than a predefined minimum width and the pixel density of the one or more image contours to be greater than the predefined pixel density.

4. The method as claimed in claim 1 , wherein processing the one or more text regions comprises:
converting, by the text detection system, the binary image to an eroded image using erosion and dilation technique;
identifying, by the text detection system, a block of text in the eroded image based on a predefined maximum height and maximum width and a predefined minimum height and minimum width and a predefined pixel density of image contours in the eroded image; and
detecting, by the text detection system, the text from the block of text using optical character recognition (OCR) technique.

5. A text detection system for detecting text in digital engineering drawings, comprising:
a processor; and
a memory communicatively coupled to the processor, wherein the memory stores processor instructions, which, on execution, causes the processor to:
receive a digital image of an engineering drawing, wherein the digital image is converted to a binary image;
detect one or more lines inclined at one or more angles, in the binary image, based on pre-defined parameters, wherein the one or more lines are removed from the binary image;
determine one or more image contours in the binary image based on connected pixels with similar characteristics;
identify one or more regions in the binary image, to be non-text regions based on a height, a width, and a pixel density, of corresponding one or more image contours and a predefined minimum height, minimum width, maximum height, maximum width, and a predefined pixel density, wherein the one or more non-text regions are removed from the binary image, wherein the pixel density is a ratio of number of white pixels and product of contour width and height;
determine one or more text regions in the binary image based on the removal of the one or more non-text regions; and
process the one or more text regions to detect text in the binary image of the engineering drawing.

6. The text detection system as claimed in claim 5 , wherein the pre-defined parameters comprise distance resolution between two lines, angle resolution of lines passing through a point, number of votes for a line, and minimum length for the line.

7. The text detection system as claimed in claim 5 , wherein the processor identifies the one or more regions to be non-text region by detecting the height of the one or more image contour corresponding to the one or more regions to be one of, greater than a predefined maximum height and lesser than a predefined minimum height, the width of the one or more image contours to be one of greater than a predefined maximum width and lesser than a predefined minimum width and the pixel density of the one or more image contours to be greater than the predefined pixel density.

8. The text detection system as claimed in claim 5 , wherein the processor processes the one or more text regions by:
converting the binary image to an eroded image using erosion and dilation technique;
identifying a block of text in the eroded image based on a predefined maximum height and maximum width, a predefined minimum height and minimum width and a predefined pixel density of image contours in the eroded image; and
detecting the text from the block of text using optical Character Recognition (OCR) technique.

9. A non-transitory computer readable medium including instructions stored thereon that when processed by at least one processor cause a text detection system to perform operations comprising:
receiving a digital image of an engineering drawing, wherein the digital image is converted to a binary image;
detecting one or more lines inclined at one or more angles, in the binary image, based on pre-defined parameters, wherein the one or more lines are removed from the binary image;
determining one or more image contours in the binary image based on connected pixels with similar characteristics;
identifying one or more regions in the binary image, to be non-text regions based on a height, a width, and a pixel density, of corresponding one or more image contours and a predefined minimum height, minimum width, maximum height, maximum width, and a predefined pixel density, wherein the one or more non-text regions are removed from the binary image, wherein the pixel density is a ratio of number of white pixels and product of contour width and height;
determining one or more text regions in the binary image based on the removal of the one or more non-text regions; and
processing the one or more text regions to detect text in the binary image of the engineering drawing.

10. The medium as claimed in claim 9 , wherein the pre-defined parameters comprise distance resolution between two lines, angle resolution of lines passing through a point, number of votes for a line, and minimum length for the line.

11. The medium as claimed in claim 9 , wherein the instruction causes the processor to identify the one or more regions to be non-text region by detecting the height of the one or more image contour corresponding to the one or more regions to be one of, greater than a predefined maximum height and lesser than a predefined minimum height, the width of the one or more image contours to be one of greater than a predefined maximum width and lesser than a predefined minimum width and the pixel density of the one or more image contours to be greater than the predefined pixel density.

12. The medium as claimed in claim 9 , wherein the instruction causes the processor to process the one or more text regions by:
converting the binary image to a

Assignees

• Wipro Ltd

Inventors

• Shanmugavel Sibi Chakravarthy

Classifications

• G06V30/422 Primary

  Technical drawings; Geographical maps · CPC title

• G06V30/414 Primary

  Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text · CPC title

• G06K9/00463 Primary

  Physics · mapped topic

• G06K9/00476

  Physics · mapped topic