Recovering depth from focus using iterative image estimation techniques
In this report we examine the possibility of using linear and nonlinear image estimation techniques to build a depth map of a three dimensional scene from a sequence of partially focused images. In particular, the techniques proposed to solve the problem of construction of a depth map are: (1) linear methods based on regularization procedures and (2) nonlinear methods based on statistical modeling. In the first case, we have implemented a matrix-oriented method to recover the point spread function (PSF) of a sequence of partially defocused images. In the second case, the chosen method has been a procedure based on image estimation by means of the EM algorithm, a well known technique in image reconstruction in medical applications. This method has been generalized to deal with optically defocused image sequences.
- Research Organization:
- Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC03-76SF00098
- OSTI ID:
- 10146758
- Report Number(s):
- LBL-35158; ON: DE94010994
- Resource Relation:
- Other Information: PBD: Sep 1993
- Country of Publication:
- United States
- Language:
- English
Similar Records
Velocity estimation for depth conversion of a 3-D data set in Troll field: A comparison of different techniques
Efficient Levenberg-Marquardt minimization of the maximum likelihood estimator for Poisson deviates