Abstract: Millimeter wave near-field synthetic aperture radar (SAR) imaging technology has important application value in the field of security inspection. However, the imaging results generally have the problems of limited resolution and artifact interference. It is easy to magnify the image artifact interference only by super-resolution reconstruction, and it will lose the target details only by artifact suppression, which is difficult to take into account the image definition and image fidelity. It is urgent to realize the collaborative optimization of super-resolution reconstruction and artifact suppression. To solve this problem, this paper proposes a near-field SAR image super-resolution reconstruction method for noise suppression and artifact elimination. Based on the real enhanced super-resolution generative adversarial network (Real-ESRGAN), this paper embeds the adaptive artifact suppression module (AASM) in the generator, and realizes the accurate identification and adaptive suppression of SAR image artifacts and noise based on the data-driven strategy; At the same time, the sparse artifact suppression loss function is constructed to effectively weaken the dark area artifact interference by suppressing the high-frequency redundant components of the image. The experimental results of ×2 and ×4 super sub tasks show that the proposed method is superior to the comparison algorithm in both subjective and objective core indexes, and can provide a feasible technical scheme for the optimization of millimeter wave near-field SAR image quality.