A contrast-based phase calibration algorithm for digital-beamforming remote sensing radars using three contrast metrics is presented. The performance of the algorithm is studied using a combination of simulated and real radar data under various conditions, and is compared with a clutter-based calibration algorithm. An analytical proof showing that maximizing the expected value of the 4-norm metric is equivalent to phase calibrating the image, is provided. We find that the clutter calibration algorithm performs best for statistically homogeneous scenes, but that the contrast calibration algorithms perform better with scenes with larger contrast ratios.