Optical image processing method, machine learning method, trained model, machine learning preprocessing method, optical image processing module, optical image processing program, and optical image processing system

1 . An optical image processing method comprising:

accepting an input of condition information indicating imaging conditions when an image of the target object is captured, wherein the condition information includes information indicating a type of photodetector used to capture an image of the target object;

acquiring an optical image obtained by capturing an image of light from a target object;

generating a noise map by:

deriving, for each pixel of the optical image, an evaluation value from the condition information and a pixel value of said each pixel on the basis of relationship data, wherein the relationship data indicates a predefined relationship between the pixel value and the evaluation value obtained by evaluating a spread of a noise value; and

associating the derived evaluation value with said each pixel of the optical image to generate the noise map; and

inputting the optical image and the noise map to a trained model built in advance through machine learning and executing image processing of removing noise from the optical image.

2 . The optical image processing method according to claim 1 , wherein acquiring of the optical image includes acquiring an optical image of a jig obtained by capturing an image of light from the jig, and

deriving of the evaluation value includes deriving the relationship data from the optical image of the jig.

3 . The optical image processing method according to claim 1 , wherein acquiring of the optical image includes acquiring a plurality of optical images captured without the target object,

deriving of the evaluation value includes deriving the relationship data from the plurality of optical images, and

the plurality of optical images are a plurality of images with imaging conditions different from each other.

4 . The optical image processing method according to claim 1 , wherein the evaluation value is a standard deviation of noise values.

5 . A machine learning method comprising:

using a structure image to which noise is added based on a predetermined noise distribution model as a training image and using the training image, a noise map generated from the training image on the basis of relationship data indicating a relationship between a pixel value and an evaluation value obtained by evaluating spread of a noise value, and noise-removed image data which is data obtained by removing noise from the training image, as training data, to build a trained model that outputs the noise-removed image data on the basis of the training image and the noise map through machine learning;

accepting an input of condition information indicating imaging conditions when an image of the target object is captured, wherein the condition information includes information indicating a type of photodetector used to capture an image of the target object;

acquiring an optical image obtained by capturing an image of light from a target object;

generating a noise map by:

deriving, for each pixel of the optical image, an evaluation value from the condition information and a pixel value of said each pixel on the basis of relationship data, wherein the relationship data indicates a predefined relationship between the pixel value and the evaluation value obtained by evaluating a spread of a noise value; and

associating the derived evaluation value with said each pixel of the optical image to generate the noise map; and

inputting the optical image and the noise map to the trained model built in advance through the machine learning and executing image processing of removing noise from the optical image.

6 . A machine learning preprocessing method in the machine learning method according to claim 5 , comprising:

generating the structure image to which noise is added based on the noise distribution model as the training image.

7 . The machine learning preprocessing method according to claim 6 ,

wherein generating of the structure image includes determining the noise distribution model from the photodetector information.

8 . The machine learning preprocessing method according to claim 7 , wherein the noise distribution model includes at least one of a normal distribution model and a Poisson distribution model.

9 . The machine learning preprocessing method according to claim 7 , wherein the noise distribution model includes a Bessel function distribution model.

10 . An optical image processing module comprising a processor configured to:

accept an input of condition information indicating imaging conditions when an image of the target object is captured, wherein the condition information includes information indicating a type of photodetector used to capture an image of the target object;

acquire an optical image obtained by capturing an image of light from a target object;

generate a noise map by:

deriving, for each pixel of the optical image, an evaluation value from the condition information and a pixel value of said each pixel on the basis of relationship data, wherein the relationship data indicates a predefined relationship between the pixel value and the evaluation value obtained by evaluating a spread of a noise value; and

associating the derived evaluation value with said each pixel of the optical image to generate the noise map; and

input the optical image and the noise map to a trained model built in advance through machine learning and execute image processing of removing noise from the optical image.

11 . The optical image processing module according to claim 10 , wherein the processor acquires an optical image of a jig obtained by capturing an image of light from the jig, and

derives the relationship data from the optical image of the jig.

12 . The optical image processing module according to claim 10 , wherein the processor acquires a plurality of optical images captured without the target object,

derives the relationship data from the plurality of optical images, and wherein the plurality of optical images are a plurality of images with imaging conditions different from each other.

13 . The optical image processing module according to claim 10 , wherein the evaluation value is a standard deviation of noise values.

14 . The optical image processing module according to claim 10 , wherein the processor uses a structure image to which noise is added based on a predetermined noise distribution model as training image and use the training image, the noise map generated from the training image on the basis of the relationship data, and noise-removed image data which is data obtained by removing noise from the training image, as training data, to build a trained model that outputs the noise-removed image data on the basis of the training image and the noise map through machine learning.

15 . A non-transitory computer-readable storage medium storing an optical image processing program causing a processor to execute a method comprising:

accepting an input of condition information indicating imaging conditions when an image of the target object is captured, wherein the condition information includes information indicating a type of photodetector used to capture an image of the target object;

acquiring an optical image obtained by capturing an image of light from a target object;

generating a noise map by:

deriving, for each pixel of the optical image, an evaluation value from the condition information and a pixel value of said each pixel on the basis of relationship data, wherein the relationship data indicates a predefined relationship between the pixel value and the evaluation value obtained by evaluating a spread of a noise value; and

associating the derived evaluation value with said each pixel of the optical image to generate the noise map; and

inputting the optical image and the noise map to a trained model built in advance through machine learning and executing image processing of removing noise from the optical image.

16 . An optical image processing system comprising:

the optical image processing module according to claim 10 ; and

an imaging device configured to acquire the optical image by capturing an image of light from the target object.