Method of extracting text from graphical images

• US 6,704,449 B1
• Filed: 10/19/2000
• Issued: 03/09/2004
• Est. Priority Date: 10/19/2000
• Status: Active Grant

• Current Assignee: National Security Agency
• Original Assignee: The United States of America As Represented By The Director National Security Agency
• Inventors: Ratner, Alan S.
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): Alavi, Amir
• Application Number:
  US09/691,168
• Time in Patent Office: 1,237 Days
• Field of Search: 382/161, 382/162, 382/164, 382/165, 382/173, 382/176, 382/181, 382/185, 382/190, 382/194, 382/198, 382/218, 382/229, 382/290, 382/292, 358/462, 358/505, 358/515, 358/520, 358/523, 358/530, 345/589, 345/590, 345/600, 345/604
• US Class Current: 382/176
• CPC Class Codes: G06V 20/62 Text, e.g. of license plates, overlay texts or captions on TV images

Abstract

A method for extracting text from images whereby the intensity of each color plane of the image is identified, the color plane with the highest intensity is converted into a black and white image, and character recognition is performed on the black and white image to determine the text of the image. The method can be used for textual word searches for Graphics Image Format (GIF) images as well as Joint Photographic Experts Group (JPEG) images, thereby providing the capability to perform text searches for images found on web pages of the World Wide Web.

Claims

1. A method for extracting text from an image having green, red, and blue color planes at least one of which includes at least one first area having an intensity level that exceeds a predetermined value, comprising the steps of:

(a) obtaining an image from an image source;

(b) separating said image into its red, green, and blue color components each having a plurality of color levels;

(c) determining the image intensity levels contained with the color planes of each of said color components, respectively;

(d) scanning on a pixel-by-pixel basis for at least one first color component the color plane having the highest intensity level;

(e) comparing the intensities of successive color pixels with said highest intensity level of said first color component, and generating corresponding black pixels for those color pixels having intensities equal to said highest intensity level, and white pixels for those color pixels having intensity levels that are not equal to said highest intensity level;

(f) comparing the total number of black pixels with the total number of white pixels;

(g) converting in the event that the number of black pixels exceeds the number of white pixels the black pixels to white pixels and the white pixels to black pixels, respectively; and

(h) recognizing the text characters of the resultant black pixels.

2. A method for extracting text as defined in claim 1, and further including the intermediate step of:

(i) repeating said scanning step, in the event that no text character is successfully recognized, for a lower intensity level of a color plane of said first color component.

3. A method for extracting text as defined in claim 2, and further including the intermediate step of:

(j) repeating said scanning steps, in the event that no text character is successfully recognized, for the highest intensity level of the color plane of a second color component.

4. A method for extracting text as defined in claim 3, and further including the intermediate step of:

(k) repeating said scanning step, in the event that no text character is successfully recognized, for a lower intensity color plane of said second color component.

5. A method for extracting text as defined in claim 1, wherein said text character recognizing step includes:

(l) comparing the image of the black and white pixels with a source of standard text values.

6. A method for extracting text as defined in claim 1, and further including the intermediate step of:

(i) comparing the intensity level of each of the color components, said first color component comprising that color component having the highest intensity level.

7. A method for extracting text as defined in claim 6, and further including the intermediate steps of:

(j) determining the number of color plane intensities for each of said color components;

(k) arranging said color plane intensities for said color components in descending order; and

(l) determining a representative sampling number of said color plane intensities for transmittal to said scanning means.