Image processing apparatus, image processing method, and computer-readable recording device

Patent metadata

Field	Value
Publication number	US-9165370-B2
Application number	US-201213396844-A
Country	US
Kind code	B2
Filing date	Feb 15, 2012
Priority date	Mar 16, 2011
Publication date	Oct 20, 2015
Grant date	Oct 20, 2015

Abstract

An image processing apparatus includes a probability value calculator that calculates, based on color feature data of a pixel included in an image, respective probabilities that the pixel belongs to a specific area and a non-specific area; a weighted edge intensity calculator that calculates a weighted edge intensity between neighboring pixels based on pixel values and color feature data of the pixel included in the image and a neighboring pixel of the pixel; an energy function creator that uses the probabilities and the weighted edge intensity to create an energy function expressed by a result of an area determination of the pixel; and an area divider that divides the image into the plurality of areas based on the energy function.

First claim

What is claimed is:

1. An image processing apparatus for processing a plurality of pixels of an image, the image being of an object comprising a first area and a second area sharing a border with the first area, the image processing apparatus comprising:
a processor comprising hardware, the processor being configured to implement:
a probability value calculator configured to calculate, based on color feature data of the plurality of pixels:
a probability that a target pixel of the plurality of pixels corresponds to the first area; and
a probability that the target pixel corresponds to the second area;

a weighted edge intensity calculator configured to calculate:
an edge intensity based on:
a pixel value of the target pixel; and
a pixel value of a neighboring pixel of the target pixel;

a weight distribution having a maximum value at a color feature data border in a color feature data space, wherein the color feature data border corresponds to a frequency where a color feature data distribution corresponding to the first area becomes comparable to a color feature data distribution corresponding to the second area;
a weight based on the weight distribution, wherein the weight corresponds to a color feature data of the target pixel and a color feature data of the neighboring pixel; and
a weighted edge intensity by applying the weight to the edge intensity;

an energy function creator configured to create an energy function based on:
the probability that the target pixel corresponds to the first area;
the probability that the target pixel corresponds to the second area; and
the weighted edge intensity; and

an area divider configured to calculate an area determination result based on optimization of the energy function, wherein the area determination result indicates whether the target pixel corresponds to the first area or the second area.

2. The image processing apparatus according to claim 1 , wherein the probability value calculator includes a distribution estimating unit configured to calculate:
a first distribution of a color feature data of a pixel of the plurality of pixel estimated to correspond to the first area;
a second distribution of a color feature data of a pixel of the plurality of pixel estimated to correspond to the second area;
the probability that the target pixel of the plurality of pixels corresponds to the first area based on the first distribution; and
the probability that the target pixel of the plurality of pixels corresponds to the second area based on the first distribution and the second distribution.

3. The image processing apparatus according to claim 2 , wherein the distribution estimating unit is configured to calculate:
a total distribution of the color feature data of the plurality of pixels; and
the first distribution and the second distribution by applying a mixed normal distribution to the total distribution.

4. The image processing apparatus according to claim 1 , wherein the weighted edge intensity calculator is configured to:
create a weight function based on the weight distribution; and
calculates the weight by using the weight function.

5. The image processing apparatus according to claim 4 , wherein the weight calculator comprises:
a distribution estimating unit configured to estimate, based on the color feature data of the plurality of pixels:
the color feature data distribution corresponding to the first area; and
the color feature data distribution corresponding to the second area;

a border detector configured to detect the color feature data border; and
a weight distribution setting unit configured to set the weight distribution in which a value becomes maximum at the color feature data border,
wherein a function corresponding to the weight distribution is treated as the weight function.

6. The image processing apparatus according to claim 5 , wherein the weight distribution setting unit is configured to set, as the weight distribution, a normal distribution whose average is the color feature data border.

7. The image processing apparatus according to claim 5 , wherein the weight distribution setting unit is configured to set, as the weight distribution, a distribution which is asymmetric with respect to the color feature data border.

8. The image processing apparatus according to claim 1 , wherein the weighted edge intensity calculator is configured to calculate the weight based on a data table which corresponds to the weight distribution.

9. The image processing apparatus according to claim 1 , wherein the weighted edge intensity calculator further comprises:
a high-intensity edge area detector configured to:
detect that the edge intensity of the target pixel is higher than a predetermined threshold;
determine that the target pixel is one of a plurality of pixels of a high-intensity edge area; and

a weight recalculator configured to recalculate a weight to be applied to the edge intensity of the plurality of pixels of the high-intensity edge area.

10. The image processing apparatus according to claim 9 , wherein the high-intensity edge area detector comprises:
an edge intensity image creator configured to create an edge intensity image based on the edge intensity between the target pixel and the neighboring pixel;
a high-intensity edge pixel detector configured to detect, from the edge intensity image, that the target pixel is one of the plurality of pixels of the high-intensity edge area; and
a labeling unit configured to label the plurality of pixels of the high-intensity edge area.

11. The image processing apparatus according to claim 9 , wherein the weight recalculator is configured to:
determine a maximum value among weights to be applied to the edge intensity of the pixels of the high-intensity edge area, and
treat the maximum value as a weight to be applied to edge intensities of the plurality of pixels of the high-intensity edge area.

12. The image processing apparatus according to claim 1 , wherein the area divider is configured to optimize the energy function by minimizing the energy function.

13. The image processing apparatus according to claim 1 , wherein the energy function creator comprises:
a first energy function creator configured to create a first energy function whose result becomes smaller as a number of pixels corresponding to a more probable area among the first area and the second area is larger; and
a second energy function creator configured to create a second energy function whose result becomes smaller as the weighted edge intensity between the target pixel and the neighboring pixel in which the border of the first area and the second area is determined to be present is higher.

14. The image processing apparatus according to claim 1 , wherein the color feature data is a hue.

15. The image processing apparatus according to claim 1 ,
wherein the image is an intraluminal image of an inside of a subject,
wherein the first area is a residue area, and
wherein the second area is a mucous membrane area comprising a groove.

16. An image processing method for processing a plurality of pixels of an image, the image being of an object comprising a first area and a second area sharing a border with the first area, the image processing method comprising:
calculating, by a probability value calculator implemented by a processing comprising hardware, based on color feature data of the plurality of pixels:
a probability that a target pixel of the plurality of pixels corresponds to the first area; and
a probability that the target pixel corresponds to the second area

Assignees

• Hirota Masashi
• Kanda Yamato
• Kitamura Makoto
• Matsuda Takehiro
• Kono Takashi
• Olympus Corp

Inventors

• Hirota Masashi
• Kanda Yamato
• Kitamura Makoto
• Matsuda Takehiro
• Kono Takashi

Classifications

• G06T7/0083 Primary

  Physics · mapped topic

• G06T2207/20076

  Probabilistic image processing · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/10068

  Endoscopic image · CPC title

• G06T7/0087

  Physics · mapped topic