METHOD OF AND APPARATUS FOR CLASSIFYING IMAGE USING HISTOGRAM ANALYSIS, AND METHOD OF AND APPARATUS FOR RECOGNIZING TEXT IMAGE USING THE HISTOGRAM ANALYSIS

• US 20100220927A1
• Filed: 03/02/2010
• Published: 09/02/2010
• Est. Priority Date: 03/02/2009
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Digital Imaging Company Limited (Samsung Electronics Co. Ltd.)
• Inventors: Kang, Tae-hoon, Kim, Seung-hun
• Granted Patent: US 8,774,514 B2
• Time in Patent Office: Days
• US Class Current: 382/170
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables
• CPC Class Codes: H04N 1/40062 Discrimination between different image types, e.g. two-tone, continuous tone

Abstract

Methods and apparatuses of classifying an image as either a text image or a non text image are provided. The method includes converting an input image into an image histogram comprising gray levels and a frequency of the gray levels; analyzing the image histogram to determine a characteristic of the image histogram; and classifying the input image as a text image or a non text image based on the characteristic. The characteristic of the image histogram may include the frequency of the gray levels, a number of peaks, a distance between peaks, and a peak width. The image may be split into blocks and a number of edge pixels determined for each block. The blocks with at least a predetermined number of edge pixels may then be classified. The entire image may then be classified based on the number of blocks determined to be a text image.

Claims

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

converting an input image into an image histogram comprising gray levels and a frequency of the gray levels;

analyzing the image histogram to determine a characteristic of the image histogram; and

classifying the input image as a text image or a non text image based on the characteristic.

2. The method of claim 1, wherein the characteristic of the image histogram comprises at least one from the group consisting of a frequency of the gray levels, a number of peaks, a distance between peaks, and a peak width.

3. The method of claim 1, wherein analyzing further comprises:

determining a number of peaks in the image histogram by counting the number of gray levels having a frequency above a predetermined frequency value and wherein classifying further comprises;

classifying the input image as a text image or a non text image based on the number of peaks.

4. The method of claim 3, wherein analyzing further comprises:

determining a number of intervals between the peaks; and

wherein classifying further comprises;

classifying the input image as a text image or a non text image based on the number of intervals.

5. The method of claim 4, wherein analyzing further comprises:

when the number of peaks is two or more, determining whether the distance between the peaks is within a predefined range; and

wherein classifying further comprises;

classifying the input image as a text image or a non text image based on whether the distance between peaks is within the predefined range.

6. The method of claim 5, wherein analyzing further comprises:

determining a width of each peak by determining the number of consecutive gray levels having a frequency above the predetermined frequency value;

determining whether each width is below a threshold value; and

wherein classifying further comprises;

classifying the input image based on whether the width of each peak is below the threshold value.

7. The method of claim 6, wherein the predetermined frequency value, the predefined range, and the threshold value are set according to a type of the input image.

8. The method of claim 2, wherein the analyzing further comprises normalizing the image histogram.

9. The method of claim 7, wherein the type of the input image comprises at least one of a text image and a photo image.

10. An apparatus for classifying an image, the apparatus comprising:

a histogram analyzing unit configured to convert an input image into an image histogram comprising gray levels and for each gray level a frequency of the gray level in the input image, and configured to calculate a characteristic of the image histogram; and

an image classifying unit configured to classify the input image based on the characteristic.

11. The apparatus of claim 10, wherein the characteristic of the image histogram comprises at least one from the group consisting of the frequency of the gray levels, a number of peaks, a distance between peaks, and a peak width.

12. The apparatus of claim 11, wherein the histogram analyzing unit comprises:

a frequency setting unit configured to set the predetermined frequency value to a predetermined value;

a peak number determining unit configured to determine a number of peaks by calculating the number of gray levels with a frequency above the predetermined frequency value;

a peak distance determining unit configured to set a first threshold range and configured to determine a distance between peaks, and configured to determine whether the distance between the peaks is within the first threshold range; and

a peak width determining unit configured to set a second threshold value and configured to determine a peak width for each of the peaks and configured to determine whether the peak width of each of the peaks is below the second threshold and configured to determine a total of the peaks widths and to determine whether the total of the peaks widths is below the second threshold value.

13. The apparatus of claim 12, wherein the image classifying unit classifies the input image according to at least one from the group consisting of the number of peaks, the distance between peaks, the peak width of each of the peaks, and the total of the peaks widths.

14. The apparatus of claim 13, wherein the predetermined frequency value, the first threshold range and the second threshold are set according to a type of the input image.

15. The apparatus of claim 12, wherein the histogram analyzing unit further comprises a normalization unit for normalizing the image histogram.

16. The apparatus of claim 14, wherein the type of the input image comprises a text image and a photograph image.

17. A method of recognizing text using histogram analysis, the method comprising:

converting an input image into a gray image and detecting pixels including edge components from the gray image;

dividing a region of the gray image into a predetermined number of blocks;

determining which blocks have a number of edge pixels above a first threshold value;

analyzing a characteristic of an image histogram for the gray image for each block having the number of edge pixels above the first threshold value and determining whether the corresponding blocks are text blocks based on the image histogram; and

recognizing the input image as a text image based on a number of corresponding blocks determined to be text blocks.

18. The method of claim 17, wherein the characteristic of the image histogram comprises at least one from the group consisting of a frequency of the gray levels, a number of peaks, a distance between peaks, and a peak width.

19. The method of claim 17, wherein recognizing further comprises:

recognizing the input image as a text image when a ratio of blocks determined to be text blocks to all blocks is above a second threshold value.

20. The method of claim 17, wherein recognizing further comprises:

recognizing the input image as a text image when a value, in which weight indicating dispersion of the text blocks is given to a ratio of blocks determined to be text blocks to all blocks, is above a third threshold value.

21. The method of claim 17, wherein determining whether the blocks are text blocks comprises:

setting a predetermined frequency value to a predetermined value;

determining a number of peaks in the image histogram by counting the number of gray levels having a gray level above the predetermined frequency value;

when the number of peaks is two or more, determining whether the distance between the peaks is within a predefined fourth threshold range;

determining a width of each peak by determining the number of consecutive gray levels having a frequency above the predetermined frequency value;

determining whether each peak width is below a fifth threshold value and determining whether a total of all peak widths is below the fifth threshold; and

determining the corresponding blocks as texts blocks based on at least one of the group consisting of;

the number of peaks, whether the distance between the peaks is within a predefined fourth threshold range, and whether each peak width is below the fifth threshold, and whether the total of all the peak widths is below the fifth threshold.

22. An apparatus for recognizing text using histogram analysis, the apparatus comprising:

an edge detecting unit configured to detect pixels that are part of edge components from a gray image;

a region dividing unit configured to divide a region of the gray image into a predetermined number of blocks;

an edge pixel determining unit configured to determine which blocks having a number of edge pixels above a first threshold value;

a histogram analyzing unit configured to analyze a characteristic of an image histogram for each block having the number of edge pixels above the first threshold value and configured to determine whether the corresponding blocks are text blocks; and

a text recognizing unit for recognizing the input image as a text image or a non text image based on a number of blocks determined to be text blocks.

23. The apparatus of claim 22, wherein the characteristic of the image histogram comprises at least one from the group consisting of a frequency of the gray levels, a number of peaks, a distance between peaks, and a peak width.

24. The apparatus of claim 22, wherein the text recognizing unit recognizes the input image as a text image when a ratio of text blocks determined to be text blocks to all blocks is above a second threshold value.

25. The apparatus of claim 22, wherein the text recognizing unit recognizes the input image as a text image when a dispersion of the blocks recognized as text blocks indicates the input image is a text image.

26. The apparatus of claim 23, wherein the histogram analyzing unit comprises:

a frequency setting unit configured to set a predetermined frequency value to a predetermined value;

a peak number determining unit configured to determine a number of peaks by calculating the number of gray levels with a frequency above the predetermined frequency value;

a peak distance determining unit configured to set a third threshold range and configured to determine whether a distance between the peaks is within the first threshold value range;

a peak width determining unit configured to set a fourth threshold value and configured to determine whether a peak width for each peak is below the fourth threshold and a total of the peak widths is below the fourth threshold value; and

an image classifying unit configured to determine that a block is a text block based on at least one from the group consisting of;

the number of the peaks, the distance between peaks, the peak width, and the total of the peak widths.