Image processing device, method and non-transitory computer readable medium
An image processing device includes multiple image processing units, each trained to accommodate a different feature possibly contained in an image, a decision unit that decides a sequence of the multiple image processing units according to the features contained in an input image, and an application unit that applies the image processing units to the input image in the sequence decided by the decision unit.
1. An image processing device comprising:
a plurality of image processing units, each trained to accommodate a different feature possibly contained in an image;
a decision unit that decides a sequence of the plurality of image processing units according to the features contained in an input image;
an application unit that applies the image processing units to the input image in the sequence decided by the decision unit; and
a training unit that, by inputting a training image and the sequence by which an image processing result for the training image becomes optimal into the decision unit, trains the decision unit to output the optimal sequence of the plurality of image processing units in a case in which the training image is input,
wherein in the case in which a plurality of the sequences for which the image processing result for the training image becomes optimal exists, the training unit trains the decision unit to prioritize associating the sequence having a greater number of different image processing units applied in parallel from among the plurality of sequences with the training image.
2. The image processing device according to claim 1 , wherein
the training unit includes a selection unit that selects the plurality of image processing units from a population of the image processing units, and
the selection unit selects the plurality of image processing units in order of a favorable image processing result when the image processing unit is applied to the training image from among the image processing units included in the population.
3. The image processing device according to claim 1 , wherein
the decision unit extracts one or more features contained in the input image, and decides the sequence on a basis of the extracted combination of one or more features.
4. The image processing device according to claim 3 , wherein
the decision unit references sequence prescription information that prescribes the sequence corresponding to a combination of one or more of the features, and decides the sequence corresponding to the combination of one or more features extracted from the input image.
5. The image processing device according to claim 4 , wherein
the sequence prescription information is information that prescribes prioritizing a selection of a sequence having a greater number of different image processing units applied in parallel.
6. The image processing device according to claim 1 , wherein
the decision unit prioritizes deciding the sequence having a greater number of different image processing units applied in parallel as the sequence to apply to the input image.
7. The image processing device according to claim 2 , wherein
the decision unit prioritizes deciding the sequence having a greater number of different image processing units applied in parallel as the sequence to apply to the input image.
8. A non-transitory computer readable medium storing a program causing a computer to execute a process for processing an image, the process comprising:
training each of a plurality of image processing units to accommodate a different feature possibly contained in an image;
deciding a sequence of the plurality of image processing units according to the features contained in an input image;
applying the image processing units to the input image in the decided sequence; and
by inputting a training image and the decided sequence by which an image processing result for the training image becomes optimal, outputting the optimal sequence of the plurality of image processing units in a case in which the training image is input,
wherein in the case in which a plurality of the sequences for which the image processing result for the training image becomes optimal exists, associating the sequence having a greater number of different image processing units applied in parallel from among the plurality of sequences with the training image is prioritized.
9. An image processing device method comprising:
training each of image processing units to accommodate a different feature possibly contained in an image;
deciding a sequence of the image processing units according to the features contained in an input image;
applying the image processing units to the input image in the decided sequence; and
by inputting a training image and the decided sequence by which an image processing result for the training image becomes optimal, outputting the optimal sequence of the plurality of image processing units in a case in which the training image is input,
wherein in the case in which a plurality of the sequences for which the image processing result for the training image becomes optimal exists, associating the sequence having a greater number of different image processing units applied in parallel from among the plurality of sequences with the training image is prioritized.