A diffractive telescope is limited by its diffraction efficiency, and its imaging quality is easily affected by non-imaging order diffracted light. To resolve this problem, this study proposes a block-matching and 3D collaborative filtering image restoration algorithm based on adaptive noise estimation. First, the noise variance of a blurred image is estimated using principal component analysis, which is then combined with the known point-spread function, and the clear image is restored using the proposed algorithm. The proposed algorithm is tested in an imaging system built from diffraction telescopes. Results of numerical simulation and measurement experiment show that the proposed algorithm can improve the modulation degree of restored images by 3.58 times compared to that of raw images and can effectively improve the details of restored images and facilitate the faint-object imaging. Therefore, the proposed algorithm provides an effective way for high-contrast imaging of faint objects using diffraction telescope systems.