Image processing method, apparatus, and program

• US 7,929,739 B2
• Filed: 04/17/2007
• Issued: 04/19/2011
• Est. Priority Date: 04/17/2006
• Status: Active Grant

• Current Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Original Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Inventors: Li, Yuanzhong
• Primary Examiner(s): Mehta; Bhavesh M
• Assistant Examiner(s): SHAH, UTPAL D
• Application Number:
  US11/785,340
• Publication Number:
  US 20080044080A1
• Time in Patent Office: 1,463 Days
• Field of Search: 382/182, 382/159, 382/180, 382/199, 382/203
• US Class Current: 382/128
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06T 2207/30056 Liver; Hepatic
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06V 10/26 Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 10/774 Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

Machine learning for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, is performed in advance to obtain an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof. An arbitrary point is set within a particular region in an input image, and a discrimination region including the particular region is set in the input image with reference to the arbitrary point. A characteristic amount is obtained from each pixel within the discrimination region, then based on the characteristic amount, an evaluation value indicating whether or not each pixel within the discrimination region is a pixel representing the contour is calculated using the evaluation function, and the contour of the particular region in the input image is determined based on the evaluation value.

Claims

1. An image processing apparatus for determining a contour of a particular region in an input image, comprising:

an evaluation function obtaining means for performing machine learning in advance for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, and obtaining an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof;

a point setting means for setting an arbitrary point within the particular region in the input image;

a region setting means for setting a discrimination region in the input image, which includes the particular region, with reference to the determined arbitrary point;

a characteristic amount obtaining means for obtaining a characteristic amount of each pixel within the determined discrimination region;

an evaluation value calculation means for calculating, using the evaluation function, an evaluation value that indicates whether or not each pixel within the discrimination region is a pixel representing the contour based on the obtained characteristic amount thereof; and

a contour determination means for determining the contour using the calculated evaluation value, wherein the characteristic amount is combinations of intensity values of a plurality of different pixels within an adjacent region to a target pixel from which the characteristic amount is to be obtained.

9. The image processing apparatus according to claim 1, wherein the evaluation function is a function capable of evaluating whether or not each pixel is a pixel representing the contour using discrimination results provided by a plurality of weak classifiers that evaluates whether or not each pixel is a pixel representing the contour.

10. The image processing apparatus according to claim 1, wherein the contour determination means is a means for determining the contour using a graph cuts algorithm.

11. The image processing apparatus according to claim 1, wherein the adjacent region has a rectangular shape.

15. The image processing apparatus according to claim 9, wherein:

the apparatus further comprises a coordinate transformation means for transforming an image within the discrimination region from an image represented in a rectangular coordinate system into an image represented in a polar coordinate system with respect to the arbitrary point; and

the characteristic amount obtaining means is a means for obtaining the characteristic amount using the image within the discrimination region coordinate-transformed and represented in the polar coordinate system.

16. The image processing apparatus according to claim 9, wherein the contour determination means is a means for determining the contour using a graph cuts algorithm.

2. An image processing apparatus for determining a contour of a particular region in an input image, comprising:

an evaluation function obtaining means for performing machine learning in advance for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, and obtaining an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof;

a point setting means for setting an arbitrary point within the particular region in the input image;

a region setting means for setting a discrimination region in the input image, which includes the particular region, with reference to the determined arbitrary point;

a characteristic amount obtaining means for obtaining a characteristic amount of each pixel within the determined discrimination region;

an evaluation value calculation means for calculating, using the evaluation function, an evaluation value that indicates whether or not each pixel within the discrimination region is a pixel representing the contour based on the obtained characteristic amount thereof; and

a contour determination means for determining the contour using the calculated evaluation value, wherein the characteristic amount is intensity information within a one-dimensional intensity profile extending from a target pixel from which the characteristic amount is to be obtained to an arbitrary direction within the discrimination region.

4. The image processing apparatus according to claim 2, wherein the evaluation function is a function capable of evaluating whether or not each pixel is a pixel representing the contour using discrimination results provided by a plurality of weak classifiers that evaluates whether or not each pixel is a pixel representing the contour.

5. The image processing apparatus according to claim 2, wherein:

the apparatus further comprises a multi-resolution image obtaining means for obtaining a plurality of images having different resolutions by multi-resolutioning an image within the discrimination region; and

the characteristic amount obtaining means and the evaluation value calculation means are means for obtaining the characteristic amount and calculating the evaluation value for each of the plurality of images respectively.

6. The image processing apparatus according to claim 2, wherein:

the apparatus further comprises a coordinate transformation means for transforming an image within the discrimination region from an image represented in a rectangular coordinate system into an image represented in a polar coordinate system with respect to the arbitrary point; and

the characteristic amount obtaining means is a means for obtaining the characteristic amount using the image within the discrimination region coordinate-transformed and represented in the polar coordinate system.

7. The image processing apparatus according to claim 2, wherein the contour determination means is a means for determining the contour using a graph cuts algorithm.

8. The image processing apparatus according to claim 2, wherein the particular region is a tumor area in a medical image.

12. The image processing apparatus according to claim 4, wherein:

the apparatus further comprises a coordinate transformation means for transforming an image within the discrimination region from an image represented in a rectangular coordinate system into an image represented in a polar coordinate system with respect to the arbitrary point; and

the characteristic amount obtaining means is a means for obtaining the characteristic amount using the image within the discrimination region coordinate-transformed and represented in the polar coordinate system.

13. The image processing apparatus according to claim 4, wherein the contour determination means is a means for determining the contour using a graph cuts algorithm.

14. The image processing apparatus according to claim 6, wherein the contour determination means is a means for determining the contour using a dynamic programming method.

17. The image processing apparatus according to claim 12, wherein the contour determination means is a means for determining the contour using a dynamic programming method.

3. An image processing apparatus for determining a contour of a particular region in an input image, comprising:

an evaluation function obtaining means for performing machine learning in advance for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, and obtaining an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof;

a point setting means for setting an arbitrary point within the particular region in the input image;

a region setting means for setting a discrimination region in the input image, which includes the particular region, with reference to the determined arbitrary point;

a characteristic amount obtaining means for obtaining a characteristic amount of each pixel within the determined discrimination region;

an evaluation value calculation means for calculating, using the evaluation function, an evaluation value that indicates whether or not each pixel within the discrimination region is a pixel representing the contour based on the obtained characteristic amount thereof; and

a contour determination means for determining the contour using the calculated evaluation value, wherein the characteristic amount is intensity information within a one-dimensional intensity profile that combines a one-dimensional intensity profile extending from the arbitrary point to the direction passing through a target pixel from which the characteristic amount is to be obtained with a one-dimensional intensity profile extending from the arbitrary point to another direction.

18. An image processing method for determining a contour of a particular region in an input image, comprising:

performing machine learning in advance for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, and obtaining an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof;

setting an arbitrary point within the particular region in the input image;

setting a discrimination region in the input image, which includes the particular region, with reference to the set arbitrary point;

obtaining a characteristic amount of each pixel within the set discrimination region;

calculating, using the evaluation function, an evaluation value that indicates whether or not each pixel within the discrimination region is a pixel representing the contour based on the obtained characteristic amount thereof; and

determining the contour using the calculated evaluation value, wherein the characteristic amount is combinations of intensity values of a plurality of different pixels within an adjacent region to a target pixel from which the characteristic amount is to be obtained.

19. An image processing method for determining a contour of a particular region in an input image, comprising:

performing machine learning in advance for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, and obtaining an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof;

setting an arbitrary point within the particular region in the input image;

setting a discrimination region in the input image, which includes the particular region, with reference to the set arbitrary point;

obtaining a characteristic amount of each pixel within the set discrimination region;

calculating, using the evaluation function, an evaluation value that indicates whether or not each pixel within the discrimination region is a pixel representing the contour based on the obtained characteristic amount thereof; and

determining the contour using the calculated evaluation value, wherein the characteristic amount is intensity information within a one-dimensional intensity profile extending from a target pixel from which the characteristic amount is to be obtained to an arbitrary direction within the discrimination region.

20. An image processing method for determining a contour of a particular region in an input image, comprising:

performing machine learning in advance for learning a characteristic amount of each pixel in a plurality of sample images, each including a particular region with a known contour, and obtaining an evaluation function capable of evaluating whether or not each pixel is a pixel representing the contour based on the characteristic amount thereof;

setting an arbitrary point within the particular region in the input image;

setting a discrimination region in the input image, which includes the particular region, with reference to the set arbitrary point;

obtaining a characteristic amount of each pixel within the set discrimination region;

calculating, using the evaluation function, an evaluation value that indicates whether or not each pixel within the discrimination region is a pixel representing the contour based on the obtained characteristic amount thereof; and

determining the contour using the calculated evaluation value, wherein the characteristic amount is intensity information within a one-dimensional intensity profile that combines a one-dimensional intensity profile extending from the arbitrary point to the direction passing through a target pixel from which the characteristic amount is to be obtained with a one-dimensional intensity profile extending from the arbitrary point to another direction.