Radiographic image processing method, trained model, radiographic image processing module, radiographic image processing program, and radiographic image processing system
A control device includes an acquisition unit configured to acquire X-ray transmission images of a jig and a target object using an image acquisition device that radiates X-rays to the target object and captures an image of the X-rays passing through the target object to acquire an X-ray transmission image, a specification unit configured to specify image characteristics of the X-ray transmission image of the jig, a selection unit configured to select a trained model on the basis of the image characteristics from a plurality of trained models constructed through machine training in advance using image data, and a processing unit configured to execute image processing for removing noise from the X-ray transmission image of the target object using the selected trained model.
1 . A radiographic image processing method comprising:
acquiring a radiographic image of a jig, the radiographic image of the jig is acquired by using a system that radiates radiation to a target object and captures an image of the radiation passing through the target object;
specifying image characteristics of the radiographic image of the jig;
selecting a trained model on a basis of the image characteristics of the radiographic image of the jig from a plurality of trained models, each of the plurality of trained models, constructed using image data of one or more training images, through a machine training in advance;
acquiring a radiographic image of the target object using the system; and
executing an image processing for removing noise from the radiographic image of the target object using the selected trained model.
2 . The radiographic image processing method according to claim 1 , wherein selecting of the trained model includes selecting the trained model by comparing the image characteristics of the radiographic image of the jig with image characteristics specified from the image data of the one or more training images used to construct the plurality of trained models.
3 . The radiographic image processing method according to claim 2 , wherein the image characteristics of the radiographic image of the jig are at least one of energy characteristics, noise characteristics, and frequency characteristics, and
selecting of the trained model includes selecting the trained model on a basis of the image characteristics of the radiographic image of the jig and image characteristics of the image data of the one or more training images used to construct the plurality of trained models, the image characteristics of the image data of the one or more training images are at least one of energy characteristics, noise characteristics, and frequency characteristics, and the image characteristics of the radiographic image of the jig and the image characteristics of the image data of the one or more training images are a same image characteristics.
4 . The radiographic image processing method according to claim 1 , wherein specifying the image characteristics of the radiographic image of the jig includes specifying image characteristics of a plurality of images obtained as a result of applying the plurality of trained models to the radiographic image of the jig, and
selecting the trained model includes selecting the trained model on a basis of the image characteristics of the plurality of images.
5 . The radiographic image processing method according to claim 4 , wherein the image characteristics of the radiographic image of the jig are resolution characteristics or a luminance to noise ratio, and
selecting the trained model includes selecting the trained model used to generate an image having relatively excellent resolution characteristics or a luminance to noise ratio.
6 . The radiographic image processing method according to claim 1 , wherein the machine training is deep learning.
7 . A radiographic image processing system comprising:
at least one processor configured to:
acquire a radiographic image of a jig and a radiographic image of a target object, the radiographic image of the jig and the radiographic image of the target object are acquired by using a system that radiates radiation to the target object and captures an image of the radiation passing through the target object to acquire the radiographic image of the target object,
specify image characteristics of the radiographic image of the jig,
select a trained model on a basis of image characteristics from a plurality of trained models constructed using image data of one or more training images, through a machine training in advance, and
execute an image processing for removing noise from the radiographic image of the target object using the selected trained model.
8 . The radiographic image processing system according to claim 7 , wherein the at least one processor selects the trained model by comparing the image characteristics of the radiographic image of the jig with image characteristics specified from the image data of the one or more training images used to construct the plurality of trained models.
9 . The radiographic image processing system according to claim 8 , wherein the image characteristics of the radiographic image of the jig are at least one of energy characteristics, noise characteristics, and frequency characteristics, and
the at least one processor selects the trained model on a basis of the image characteristics of the radiographic image of the jig and image characteristics of the image data of the one or more training images used to construct the plurality of trained models, the image characteristics of the image data of the one or more training images are at least one of energy characteristics, noise characteristics, and frequency characteristics, and the image characteristics of the radiographic image of the jig and the image characteristics of the image data of the one or more training images are a same image characteristics.
10 . The radiographic image processing system according to claim 7 , wherein the at least one processor specifies image characteristics of a plurality of images obtained as a result of applying the plurality of trained models to the radiographic image of the jig, and
the at least one processor selects the trained model on a basis of the image characteristics of the plurality of images.
11 . The radiographic image processing system according to claim 10 , wherein the image characteristics of the radiographic image of the jig are resolution characteristics or a luminance to noise ratio, and
the at least one processor selects the trained model used to generate an image having relatively excellent resolution characteristics or a luminance to noise ratio.
12 . The radiographic image processing system according to claim 7 , wherein the machine training is deep learning.
13 . The radiographic image processing system according to claim 7 , further comprising:
a source configured to radiate radiation to the target object; and
a detector configured to capture the image of the radiation passing through the target object and acquire the radiographic image of the jig and the radiographic image of the target object.
14 . A non-transitory computer-readable media storing a radiographic image processing program, which, when executed by a processor, causes the processor to execute a method comprising:
acquiring a radiographic image of a jig and a radiographic image of a target object using a system that radiates radiation to the target object and captures an image of the radiation passing through the target object to acquire the radiographic image of the target object;
specifying image characteristics of the radiographic image of the jig;
selecting a trained model on a basis of image characteristics from a plurality of trained models constructed using image data of one or more training images, through a machine training in advance; and
executing an image processing for removing noise from the radiographic image of the target object using the selected trained model.