Enhancing videos or images on resolution, frame rate, dynamic range, color gamut, noise, or scratches removing have attracted the interest of experts. Efficient image quality assessment methods for these image enhancement algorithms should be available. Many image quality assessment metrics work well on the general image distortion assessment databases. However, they may not be able to distinguish artifacts caused by the image enhancement algorithms. In this paper, we study the reliability of state-of-the-art prominent full-reference and no-reference quality assessment metrics on image resolution enhancement artifacts. Firstly, we construct a 4K resolution database containing 1152 enhanced images with different superresolution distortion types and different distortion levels. Then, a subjective study with over 20000 human judgments is organized to reach reliable references from human visual perception. The performance comparison experiments on the proposed database show that IFC, FSIM, and DISTS are the top 3 perceptualconsistent metrics for image super-resolution artifacts assessment.