Image fusion method, model training method, and related apparatuses

An apparatus for fusing at least one image, the apparatus comprising:

• a memory storing instructions;
• and
• a processor in communication with the memory, wherein, when the processor executes the instructions, the processor is configured to cause the apparats to:
• obtain a first to-be-fused image and a second to-be-fused image, the first to-be-fused image comprising a plurality of first regions, and the second to-be-fused image comprising a plurality of second regions,
• obtain a first feature set according to the first to-be-fused image and obtain a second feature set according to the second to-be-fused image, the first feature set comprising a plurality of first encoding features and the second feature set comprising a plurality of second encoding features, the plurality of the first encoding features being in a one-to-one correspondence with the plurality of the first regions and the plurality of the second encoding features being in a one-to-one correspondence with the plurality of the second regions,
• perform first fusion processing on the first to-be-fused image and the second to-be-fused image by using a shape fusion network model to obtain a third to-be-fused image, the third to-be-fused image comprising at least one first encoding feature and at least one second encoding feature, and
• perform second fusion processing on the third to-be-fused image and the first to-be-fused image by using a condition fusion network model to obtain a target fused image,
• wherein, before the processor is configured to cause the apparats to perform the first fusion processing on the first to-be-fused image and the second to-be-fused image by using the shape fusion network model to obtain the third to-be-fused image, the processor is configured to cause the apparats to:
• determine a to-be-replaced region in the first to-be-fused image, the to-be-replaced region being one region in the plurality of first regions;
• and
• obtain a first to-be-replaced encoding feature in the first to-be-fused image according to the to-be-replaced region and obtain a second to-be-replaced encoding feature in the second to-be-fused image according to the to-be-replaced region, the first to-be-replaced encoding feature being one encoding feature in the plurality of first encoding features, and the second to-be-replaced encoding feature being one encoding feature in the plurality of second encoding features.

A non-transitory computer-readable storage medium storing computer readable instructions, wherein, the computer readable instructions, when executed by a processor, are configured to cause the processor to perform:

• obtaining a first to-be-fused image and a second to-be-fused image, the first to-be-fused image comprising a plurality of first regions, and the second to-be-fused image comprising a plurality of second regions;
• obtaining a first feature set according to the first to-be-fused image and obtaining a second feature set according to the second to-be-fused image, the first feature set comprising a plurality of first encoding features and the second feature set comprising a plurality of second encoding features, the plurality of the first encoding features being in a one-to-one correspondence with the plurality of the first regions and the plurality of the second encoding features being in a one-to-one correspondence with the plurality of the second regions;
• performing first fusion processing on the first to-be-fused image and the second to-be-fused image by using a shape fusion network model to obtain a third to-be-fused image, the third to-be-fused image comprising at least one first encoding feature and at least one second encoding feature;
• and
• performing second fusion processing on the third to-be-fused image and the first to-be-fused image by using a condition fusion network model to obtain a target fused image,
• wherein, before the computer readable instructions are configured to cause the processor to perform the first fusion processing on the first to-be-fused image and the second to-be-fused image by using the shape fusion network model to obtain the third to-be-fused image, the computer readable instructions are configured to cause the processor to perform:
• determining a to-be-replaced region in the first to-be-fused image, the to-be-replaced region being one region in the plurality of first regions;
• and
• obtaining a first to-be-replaced encoding feature in the first to-be-fused image according to the to-be-replaced region and obtaining a second to-be-replaced encoding feature in the second to-be-fused image according to the to-be-replaced region, the first to-be-replaced encoding feature being one encoding feature in the plurality of first encoding features, and the second to-be-replaced encoding feature being one encoding feature in the plurality of second encoding features.