Method for image binarization

• US 6,351,566 B1
• Filed: 03/02/2000
• Issued: 02/26/2002
• Est. Priority Date: 03/02/2000
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines SA (International Business Machines Corporation)
• Inventors: Zlotnick, Aviad
• Primary Examiner(s): LEE, TOMMY D
• Application Number:
  US09/519,446
• Time in Patent Office: 726 Days
• Field of Search: 358/455, 358/456, 358/465, 358/466, 382/237, 382/266, 382/270
• US Class Current: 382/237
• CPC Class Codes: G06T 2207/10008 from scanner, fax or copier
• CPC Class Codes: G06T 2207/30176 Document
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/162 Quantising the image signal
• CPC Class Codes: H04N 1/403 Discrimination between the two tones in the picture signal of a two-tone original

Abstract

A method for binarizing a gray-level input image including a plurality of pixels having respective gray-level values. A lower threshold and an upper threshold are determined. A first binary value is assigned to the pixels in the gray-level image having gray-level values above the upper threshold, and a second binary value is assigned to the pixels in the gray-level image having gray-level values below the lower threshold. The pixels in an intermediate group having gray-level values between the lower and upper thresholds are processed so as to determine optimal assignments of the pixels in the intermediate group to the first and second binary values.

Claims

1. A method for image binarization, comprising:

receiving a gray-level input image comprising a plurality of pixels having respective gray-level values;

analyzing variations among the gray-level values of the pixels in the input image so as to determine image statistics with respect to gaps between the gray-level values of neighboring pixels;

responsive to the image statistics, determining a lower threshold and an upper threshold, which is greater than the lower threshold by a selected difference;

assigning a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold; and

processing the pixels in an intermediate group having gray-level values between the lower and upper thresholds so as to determine optimal assignments of the pixels in the intermediate group to the first and second binary values.

2. A method according to claim 1, wherein analyzing the variations among the gray-level values comprises finding edges in the input image, and wherein determining the thresholds comprises selecting the thresholds so as to preserve the edges in an output image made up of the assigned binary values.

3. A method according to claim 1, wherein determining the thresholds comprises selecting the thresholds so as to preserve the gaps that are significant in preference to the gaps that are not significant in an output image made up of the assigned binary values.

4. A method according to claim 3, wherein selecting the thresholds comprises defining the gaps that are significant as those whose absolute magnitude is greater than the selected difference between the upper and lower thresholds.

5. A method according to claim 1, wherein determining the thresholds comprises selecting the thresholds so as to preserve edge information in an output image made up of the assigned binary values.

6. A method according to claim 5, wherein selecting the thresholds comprises choosing the thresholds substantially without dependence on the type of image feature to which the information belongs.

7. A method according to claim 5, wherein selecting the thresholds comprises finding an optimal average value of the upper and lower thresholds and finding an optimal value of the selected difference between the thresholds.

8. A method according to claim 1, wherein processing the pixels in the intermediate group comprises analyzing variations among the gray-level values of the pixels in the input image and determining the assignments of the pixels to the first and second binary values responsive to the analyzed variations.

9. A method according to claim 8, wherein determining the assignments responsive to the analyzed variations comprises finding a significant difference between the gray-level value of one of the pixels and the gray-level values of other pixels in its neighborhood, and assigning the pixel to the first or second binary value responsive to the difference.

10. A method according to claim 1, wherein processing the pixels in the intermediate group comprises applying a binarization method optimized for text to determine the optimal assignments of the pixels in the intermediate group.

11. A method according to claim 1, and comprising outputting a binary image made up of the assigned binary values of the pixels.

12. A method for image binarization, comprising:

receiving a gray-level input image comprising a plurality of pixels having respective gray-level values;

analyzing variations among the gray-level values of the pixels in the input image so as to find gaps between the gray-level values of neighboring pixels;

responsive to the gaps that are found, determining a lower threshold and an upper threshold, which is greater than the lower threshold by a selected difference;

assigning a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold; and

processing the pixels in an intermediate group having gray-level values between the lower and upper thresholds so as to determine optimal assignments of the pixels in the intermediate group to the first and second binary values,

wherein the thresholds are selected so as to preserve the gaps that are significant in preference to the gaps that are not significant in an output image made up of the assigned binary values, wherein the significant gaps are defined as those whose absolute magnitude is greater than the selected difference between the upper and lower threshold, and

wherein the thresholds are selected so as to maximize a merit score computed for multiple different pairs of upper and lower thresholds, wherein the score correlates positively with the number of significant gaps preserved in the output image by the selected thresholds, and correlates negatively with the number of gaps that are not significant that are preserved and the number of significant gaps that are not preserved in the output image by the selected thresholds.

13. A method for processing a gray-level input image, which includes a plurality of pixels having respective gray-level values, the method comprising:

analyzing variations among the gray-level values of the pixels in the input image;

responsive to the analyzed variations, determining a lower threshold and an upper threshold, which is greater than the lower threshold by a selected gap size;

assigning a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold; and

generating an output image, in which the pixels assigned the first and second binary values are represented by their respective binary values, and the pixels in an intermediate group having gray-level values between the lower and upper thresholds are represented by their respective gray-level values.

14. A method according to claim 13, wherein determining the thresholds comprises selecting the thresholds so as to preserve edge information in the output image.

15. A method according to claim 13, wherein generating the output image comprises displaying the output image.

16. A method according to claim 13, wherein generating the output image comprises storing the output image in a memory.

17. Apparatus for image binarization, comprising an image processor, which is coupled to receive a gray-level input image comprising a plurality of pixels having respective gray-level values, and which is adapted to analyze variations among the gray-level values of the pixels in the input image so as to determine image statistics with respect to gaps between the gray-level values of neighboring pixels, and responsive to the statistics, to determine a lower threshold and an upper threshold, which is greater than the lower threshold by a selected difference, to assign a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold, and to process the pixels in an intermediate group having gray-level values between the lower and upper thresholds so as to determine optimal assignments of the pixels in the intermediate group to the first and second binary values.

18. Apparatus according to claim 17, wherein the processor is adapted to determine the assignments of the pixels in the intermediate group to the first and second binary values responsive to the analyzed variations.

19. Apparatus according to claim 17, wherein the processor is adapted to select the thresholds so as to preserve edge information in an output image made up of the assigned binary values, substantially without dependence on the type of image feature to which the information belongs.

20. Apparatus according to claim 17, wherein the processor is adapted to apply a binarization method optimized for text to determine the optimal assignments of the pixels in the intermediate group.

21. Apparatus according to claim 17, wherein the processor is adapted to output a binary image made up of the assigned binary values of the pixels.

22. Apparatus for processing a gray-level input image, which includes a plurality of pixels having respective gray-level values, the apparatus comprising an image processor, which is adapted to analyze variations among the gray-level values of the pixels in the input image and, responsive to the analyzed variations, to determine a lower threshold and an upper threshold, which is greater than the lower threshold by a selected gap size, and to assign a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold, thus to generate an output image, in which the pixels assigned the first and second binary values are represented by their respective binary values, and the pixels in an intermediate group having gray-level values between the lower and upper thresholds are represented by their respective gray-level values.

23. Apparatus according to claim 22, and comprising a display, which is coupled to the processor so as to receive and display the output image.

24. Apparatus according to claim 22, and comprising a storage memory, which is coupled to the processor so as to receive and store the output image.

25. A computer software product for processing an input image, comprising a computer-readable medium having program instructions stored therein, which instructions, when read by a computer, cause the computer to receive a gray-level input image comprising a plurality of pixels having respective gray-level values, to analyze variations among the gray-level values of the pixels in the input image so as to determine image statistics with respect to gaps between the gray-level values of neighboring pixels, and responsive to the statistics, to determine a lower threshold and an upper threshold, which is greater than the lower threshold by a selected difference, to assign a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold, and to process the pixels in an intermediate group having gray-level values between the lower and upper thresholds so as to determine optimal assignments of the pixels in the intermediate group to the first and second binary values.

26. A product according to claim 25, wherein the instructions further cause the computer to output a binary image made up of the assigned binary values of the pixels.

27. A computer software product for processing a gray-level input image, which includes a plurality of pixels having respective gray-level values, the product comprising a computer-readable medium having program instructions stored therein, which instructions, when read by a computer, cause the computer to analyze variations among the gray-level values of the pixels in the input image and, responsive to the analyzed variations, to determine a lower threshold and an upper threshold, which is greater than the lower threshold by a selected gap size, to assign a first binary value to the pixels in the gray-level image having gray-level values above the upper threshold and a second binary value to the pixels in the gray-level image having gray-level values below the lower threshold, and to generate an output image, in which the pixels assigned the first and second binary values are represented by their respective binary values, and the pixels in an intermediate group having gray-level values between the lower and upper thresholds are represented by their respective gray-level values.