Training method of image correction model and color correction method of slice image

The invention relates to the field of pathological section image processing, in particular to a training method of an image correction model and a color correction method of a section image, and the method comprises the steps: A1, constructing the image correction model which comprises a generator and a discriminator; a2, generating a loss function of the image correction model; a3, inputting the slice image into a generator, generating a target image according to the slice image, and evaluating the target image to generate a first score; a4, generating a reconstructed image according to the target image, and evaluating the reconstructed image to generate a second score; the image loss is quantized by adopting a loss function; and A5, adjusting the generator and the discriminator, and then returning to the step A3 until the training is completed. The method has the beneficial effects that by training the image correction model generated based on the cyclic generative adversarial network, the slice image can b