Image processing apparatus, image processing method, and program

1. An image processing apparatus comprising:

an image correction unit that repeatedly performs an image correction process using a plurality of processing units in at least two stages which include first-stage to final-stage processing units, wherein

the image correction unit inputs a low-quality image which is an image to be corrected and a high-quality image which is a reference image,

each of the plurality of processing units in each stage performs a correction process for the low-quality image, using a class correspondence correction coefficient classified in accordance with a class corresponding to a feature amount extracted from the high-quality image or a degraded image of the high-quality image,

the class correspondence correction coefficient is generated by a learning process,

the first-stage processing unit performs the correction process for the low-quality image, using a class correspondence correction coefficient corresponding to a feature amount extracted from a degraded image of the high-quality image having a degradation level that is substantially equal to a degradation level of the low-quality image which is the image to be corrected, and

wherein the image correction unit and the plurality of processing units are each implemented via at least one processor.

2. The image processing apparatus according to claim 1 , wherein

among the plurality of processing units in each stage, a processing unit in a previous stage performs the correction process for the low-quality image, using a class correspondence correction coefficient corresponding to a feature amount extracted from a degraded image of the high-quality image which has a higher degradation level than that in a processing unit in a subsequent stage.

3. The image processing apparatus according to claim 1 , wherein

the class correspondence correction coefficient is generated by the learning process which is performed in advance on a basis of a sample image.

4. The image processing apparatus according to claim 3 , wherein

the class correspondence correction coefficient corresponds to a feature amount extracted from the sample image or a degraded image of the sample image, and

among the plurality of processing units in each stage, a processing unit in a previous stage performs the correction process, using a class correspondence correction coefficient corresponding to a feature amount extracted from an image which has a higher degradation level than that in a processing unit in a subsequent stage.

5. The image processing apparatus according to claim 1 , wherein

among the plurality of processing units in each stage, a processing unit in a previous stage performs the correction process for the low-quality image, using a class correspondence correction coefficient corresponding to a feature amount extracted from an image which has a lower resolution than that in a processing unit in a subsequent stage.

6. The image processing apparatus according to claim 1 , wherein

the class correspondence correction coefficient is a correction coefficient associated with a set class based on a combination of a feature amount of an image with a high degradation level and a feature amount of an image with a low degradation level.

7. The image processing apparatus according to claim 1 , wherein

the class correspondence correction coefficient includes a multiplication coefficient corresponding to a reference pixel applied to a filter that calculates a value of a pixel to be corrected.

8. The image processing apparatus according to claim 1 , wherein

each of the plurality of processing units in each stage includes a tap selection unit that sets a reference pixel range which is referred to in a case in which a correction value of a pixel to be corrected in the low-quality image which is the image to be corrected is calculated.

9. The image processing apparatus according to claim 8 , wherein

the tap selection unit sets the reference pixel range in accordance with a feature amount of a pixel region including the pixel to be corrected.

10. The image processing apparatus according to claim 1 , wherein

the feature amount is any one of:

(a) brightness distribution information;

(b) blurred state information; and

(c) noise information.

11. The image processing apparatus according to claim 1 , wherein

the low-quality image which is the image to be corrected is a far-infrared image or a fluorescent image.

12. The image processing apparatus according to claim 1 , wherein

the high-quality image which is the reference image is a visible image.

13. An image processing apparatus comprising:

an image correction unit that repeatedly performs an image correction process using a plurality of processing units in at least two stages which include first-stage to final-stage processing units;

a high-quality imaging unit that performs a process of capturing a visible image; and

a low-quality imaging unit that performs a process of capturing a far-infrared image or a fluorescent image, wherein

the image correction unit receives a high-quality image and a low-quality image captured by the high-quality imaging unit and the low-quality imaging unit and performs the correction process for the low-quality image,

the low-quality image is an image to be corrected and the high-quality image is a reference image,

each of the plurality of processing units in each stage performs a correction process for the low-quality image, using a class correspondence correction coefficient classified in accordance with a class corresponding to a feature amount extracted from the high-quality image or a degraded image of the high-quality image,

the class correspondence correction coefficient is generated by a learning process, and

wherein the image correction unit, the plurality of processing units, the high-quality imaging unit, and the low-quality imaging unit are each implemented via at least one processor.

14. An image processing method performed in an image processing apparatus that includes an image correction unit that repeatedly performs an image correction process using a plurality of processing units in at least two stages which include first-stage to final-stage processing units, the image correction unit and the plurality of processing units each being implemented via at least one processor, the method comprising:

inputting, by the image correction unit, a low-quality image which is an image to be corrected and a high-quality image which is a reference image; and

performing, by each of the plurality of processing units in each stage, a correction process for the low-quality image, using a class correspondence correction coefficient classified in accordance with a class corresponding to a feature amount extracted from the high-quality image or a degraded image of the high-quality image, wherein

the class correspondence correction coefficient used in the correction process is generated by a learning process, and

the first-stage processing unit performs the correction process for the low-quality image, using a class correspondence correction coefficient corresponding to a feature amount extracted from a degraded image of the high-quality image having a degradation level that is substantially equal to a degradation level of the low-quality image which is the image to be corrected.

15. A non-transitory computer-readable medium having embodied thereon a program, which when executed by one or more processors of an image processing apparatus that also includes an image correction unit that repeatedly performs an image correction process using a plurality of processing units in at least two stages which include first-stage to final-stage processing units whereby the image correction unit and the plurality of processing units are each implemented via at least one processor, causes the image processing apparatus to execute an image processing method, the method comprising:

inputting, by the image correction unit, a low-quality image which is an image to be corrected and a high-quality image which is a reference image; and

performing, by each of the plurality of processing units in each stage, a correction process for the low-quality image, using a class correspondence correction coefficient classified in accordance with a class corresponding to a feature amount extracted from the high-quality image or a degraded image of the high-quality image, wherein

the class correspondence correction coefficient used in the correction process is generated by a learning process, and

the first-stage processing unit performs the correction process for the low-quality image, using a class correspondence correction coefficient corresponding to a feature amount extracted from a degraded image of the high-quality image having a degradation level that is substantially equal to a degradation level of the low-quality image which is the image to be corrected.