Perception of depth is a central problem in machine vision. Stereo is an attractive technique for depth perception because, compared to monocular techniques, it leads to more direct, unambiguous, and quantitative depth measurements, and unlike such “active” approaches as radar and laser ranging, it is suitable in almost all application domains. We broadly define computational stereo as the recovery of the three-dimensional characteristics of a scene from multiple images taken from different points of view. The first part of the paper identifies and discusses each of the functional components of the computational stereo paradigm: image acquisition, camera modeling, feature acquisition, matching, depth determination, and interpolation. The second part discusses the criteria that are important for evaluating the effectiveness of various computational stereo techniques. The third part surveys a representative sampling of computational stereo research.