Cross-Modality Image Transformation Using Generative Adversarial Network

A cross-modality image transformation method using the Generative Adversarial Network (GAN) is proposed in this paper. First, the preprocessing methods are performed to the original image data. The computational steps include adjusting image size and suppressing image noises. Second, a kind of GAN is constructed. Both the generator network and discriminator network are created. Third, the GAN is trained and the back propagation optimization is performed continuously. Finally, the trained GAN can be used for data enhancement. To test the effects of computational steps above, the image transformation from the visible light image to other noise-contaminated image is performed. Lots of experiments have illustrated the correctness of our method.