About data preprocessing

[HWSQHL]

In Section 4.1 of the paper， “For training a model on samples in the image domain, we normalize the pixel values to the range [−1, 1]. In the frequency domain, we first convert the images using DCT, then log-scale the coefficients and, finally, normalize them by removing the mean and scaling to unit variance.”
How do you process images? Can you give me a code to process it.I want to use pytorch to realize it.