Image classification using concentration ratio

• US 7,532,755 B2
• Filed: 01/22/2004
• Issued: 05/12/2009
• Est. Priority Date: 01/22/2004
• Status: Expired due to Fees

• Current Assignee: Lexmark International Incorporated (Ninestar Corporation)
• Original Assignee: Lexmark International Incorporated (Ninestar Corporation)
• Inventors: Thakur, Khageshwar
• Primary Examiner(s): Bali; Vikkram
• Assistant Examiner(s): ZARKA, DAVID PETER
• Application Number:
  US10/763,129
• Publication Number:
  US 20050163385A1
• Time in Patent Office: 1,937 Days
• Field of Search: 382/170
• US Class Current: 382/170
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns

Abstract

Methods and systems of classifying an image by use of a concentration ratio that indicates the smoothness of the distribution of a population of elements in the image or portions thereof. Alternate embodiments may include determining the concentration ratio based on luminance components, grayscale components or a histogram, use of thresholds for determing the image classifications such as text, graphic or photographic, and use of a training process for creating the thresholds that includes analyzing a set of images having known classifications.

Claims

1. A method of classifying an image, the method comprising:

obtaining an image;

determining one or more classification thresholds;

determining a concentration ratio for the image;

comparing the concentration ratio to at least one of the one or more classification thresholds; and

classifying, using a processor, the image based on the comparison of the concentration ratio to at least one of the one or more classification thresholds, wherein determining the concentration ratio for the image includes determining the concentration ratio according to the following

$\mathrm{CR}={\left(\underset{L}{\sum <br><br>}<br><br>P_L\right)}^n/\left(\underset{L}{\sum <br><br>}<br><br>P_L^{<br><br>n}\right)$

where CR is a concentration ratio, n is greater than 1, and P_L is a population at a level L.

2. A method as claimed in claim 1 wherein n is an even integer.

3. A computer readable medium having stored thereon instructions executable by an image classifying processor to obtain an image, obtain one or more classification thresholds, determine a concentration ratio for the image, compare the concentration ratio to at least one of the one or more classification thresholds, and classify the image based on the comparison of the concentration ratio to at least one of the one or more classification thresholds, wherein the concentration ratio for the image is determined according to the following:

$\mathrm{CR}={\left(\underset{L}{\sum <br><br>}<br><br>P_L\right)}^n/\left(\underset{L}{\sum <br><br>}<br><br>P_L^{<br><br>n}\right)$

where CR is a concentration ratio, n is greater than 1, and P_L is a population at a level L.

4. A method of processing an image, the method comprising:

capturing an image of an object;

classifying the image in a class using a concentration ratio;

using the class to modify the operation of an image capturing device; and

applying, using a processor, controlled, equalization to an image generated by the image capture device, where the controlled, histogram equalization uses a concentration ratio that indicates a relative level of smoothness of a distribution of a population of elements in the image,

wherein the concentration ratio is determined according to the following;

$\mathrm{CR}={\left(\underset{L}{\sum <br><br>}<br><br>P_L\right)}^n/\left(\underset{L}{\sum <br><br>}<br><br>P_L^{<br><br>n}\right)$

where CR is the concentration ratio, n is greater than 1, and P_L is a population at a level L.

5. An image processing system comprising:

an image capture device;

an image classifier coupled to the image capture device in a feedback loop; and

a controlled, equalization processor coupled to the image capture device, that uses a concentration ratio that indicates a relative level of smoothness of a distribution of a population of elements in the image, wherein the processor is configured to determine the concentration ratio for the image according to the following;

$\mathrm{CR}={\left(\underset{L}{\sum <br><br>}<br><br>P_L\right)}^n/\left(\underset{L}{\sum <br><br>}<br><br>P_L^{<br><br>n}\right)$

where CR is the concentration ratio, n is greater than 1, and P_L is a population at a level L.

6. An image processing system comprising:

an image capture device configured to capture an image; and

an image classifier coupled to the image capture device in a feedback loop, the image classifier configured to determine a concentration ratio for the image that indicates a relative level of smoothness of a distribution of a population of elements in the image, compare the concentration ratio to at least one or more classification thresholds, and classify the image based on the comparison of the concentration ratio to at least one of the one or more classification thresholds, wherein the image classifier is configured to determine the concentration ratio for the image according to the following;

$\mathrm{CR}={\left(\underset{L}{\sum <br><br>}<br><br>P_L\right)}^n/\left(\underset{L}{\sum <br><br>}<br><br>P_L^{<br><br>n}\right)$

where CR is the concentration ratio, n is greater than 1, and P_L is a population at a level L.