*2.3. Architecture*

As shown in Figure 1, the encoder–decoder-discriminator pair constitutes the SDT-GAN model. Considering that the intrinsic information of an image is shared among multiple spectral domains, the output matrix dimensions of the encoder of all spectraldomain models are consistent.

In the process of image translation from source spectral domain *χi* to target spectral domain *χj*, the source-domain encoder *Ei* is selected from the encoder library, and the target-domain decoder *Gj* is selected from the decoder library. The encoder *Ei* maps the input matrix to the shared latent code s, and the decoder *Gj* reconstructs the target spectral-domain image from the latent code s. Then, the adversarial loss is calculated by the target-domain discriminator *Dj*. Since the latent code is shared in each spectral domain, the latent code generated by the source spectral-domain encoding can be decoded into multiple codes in the target spectral domain. Meanwhile, the input matrices need to be preprocessed, including the source spectral image matrix and the condition information matrix after feature embedding.
