文档介绍：Abstract
Image is one of the major media for information transmission of human being, however, the information contained in two-dimensional image can not meet people’s needs in general. Therefore, restoring the three-dimensional information in two-dimensional image is particularly important and is one of the hotpots puter vision.
The most important thing in three-dimensional reconstruction is to restore the depth of scenery. Recently, most depth recovery algorithms puter vision are based on focus images, such as recovery algorithms based on stereo vision, motion, and so on. Compared with other algorithms, these methods are more accurate. But they must firstly solve plex problem of choosing and matching the character points. The image recovery algorithm based on defocus images can avoid this problem. This paper mainly focused on the research of depth recovery method based on defocus images.
In this paper, we firstly summarized mon depth recovery methods. Depth recovery methods are mainly classified into two categories puter vision. One is single view depth recovery methods using one image or a number of images taken by a single view to restore depth. The other is multiview depth recovery methods using one image or a number of images taken by multiview view to restore depth. Then we introduced the research status of the depth recovery based on defocus the same time we illuminated the defocus imaging models and analyze some basic principles of depth recovery methods which are based on defocused bined with imaging models and the corresponding optical knowledge. What’s more, we also introduced the earliest and the classical Pentland defocus depth recovery methods, Polynomial method based on spatial domain and Minimize the cost function method made in recent years and summarized the reasons that bring errors in these methods.
Minimize the cost function method is on behalf of the latest developments and
depth recovery method based on geometrical optics is on