Abstract

There are many cues to recovery scene depth such as parallax, perspective and defocused information. Depth from defocus (DFD) and depth from focus (DFF) are the classical methods of a monocular vision which is based on defocused images, and the critical issue is the focus or defocus measure. The techniques of DFD and DFF are attractive because they can avoid the inaccurate matching in stereo vision and achieve the high precision of scene depth map. We applied the feature density to indicate the focus degree of an object point in the focal stack. The focus measure leads to establish computational imaging, including depth estimation and all- in-focus imaging algorithms.

© 2018 The Japan Society of Applied Physics