The invention discloses an image denoising method based on self-adaptive weighted total variation. The method includes the following steps: step 1, establishing a weighted-total-variation image denoising model; step 2, normalizing an image; step 3, constructing a matrix pair of (u, v), and carrying out initial value assignment thereon; step 4, starting a k-th iteration, and generating values of weighting parameters of W<1> and W<2> in a self-adaptive manner according to values of a matrix pair of (u, v); step 5, obtaining a denoised image of x of the k-th iteration; and step6, judging whether an iteration number of k reaches a set iteration number of N, if the iteration number of k does not reach the set iteration number of N, utilizing a gradient projection algorithm to obtain a matrix pair of (u, v), letting k = k + 1, and re-entering the step 4, and obtaining a final denoised image of x if the iteration number of k reaches the set iteration number of N.