Depth from Defocus and Coded Apertures for 3D Scene Sensing

Abstract

Depth from defocus (DfD) is one of the popular passive depth sensing approaches in computer vision. Usually, it utilizes the defocus blur depth cue encoded in images, captured by a single photographic camera. Since the defocus blur is characterized by the aperture shape of the camera, it can actually be structured by inserting a coded mask in the camera aperture position. Therefore, by employing optimized masks in such coded aperture cameras, it is possible to improve the performance of DfD approaches. DfD and coded aperture approaches constitute the main theme of this chapter. Noting, however, that the stereopsis cue is actually the most widely utilized depth cue in computer vision, joint utilization of the defocus blur cue, and the stereopsis cue is another interesting topic, which is therefore also included in the discussion.