Error Bounds in Depth from Defocus

Depth from defocus (DFD) involves estimating the relative blur between a pair of defocused images of a scene captured with di erent lens settings. When a priori information about the scene is available, it is possible to estimate the depth even from a single image. However, experimental studies indicate that the depth estimate improves with multiple observations. In this paper, we provide a mathematical underpinning to this evidence by deriving and comparing the theoretical bounds for the error in the estimate of blur corresponding to the case of a single image, and a pair of defocused images, respectively. A new theorem is proposed which proves that the Cram er-Rao bound on the variance of the error in the estimate of blur, decreases with an increase in the number of observations. The di erence in the bounds turns out to be a function of the relative blurring between the observations. Results on synthetic as well as real data are given to validate the claim.

[1]  N. Kiryati,et al.  Depth from Defocus vs. Stereo: How Different Really Are They? , 1998, Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170).

[2]  Djemel Ziou,et al.  Passive depth from defocus using a spatial domain approach , 1998, Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271).

[3]  Shree K. Nayar,et al.  Minimal operator set for passive depth from defocus , 1996, Proceedings CVPR IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[4]  Wilson S. Geisler,et al.  Maximum-likelihood depth-from-defocus for active vision , 1995, Proceedings 1995 IEEE/RSJ International Conference on Intelligent Robots and Systems. Human Robot Interaction and Cooperative Robots.

[5]  Mats Gökstorp,et al.  Computing depth from out-of-focus blur using a local frequency representation , 1994, International Conference on Pattern Recognition.

[6]  Murali Subbarao,et al.  Depth from defocus by changing camera aperture: a spatial domain approach , 1993, Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[7]  Peter Lawrence,et al.  An Investigation of Methods for Determining Depth from Focus , 1993, IEEE Trans. Pattern Anal. Mach. Intell..

[8]  Shang-Hong Lai,et al.  A Generalized Depth Estimation Algorithm with a Single Image , 1992, IEEE Trans. Pattern Anal. Mach. Intell..

[9]  Eric Paul Krotkov,et al.  Active Computer Vision by Cooperative Focus and Stereo , 1989, Springer Series in Perception Engineering.

[10]  Murali Subbarao.  Parallel Depth Recovery By Changing Camera Parameters , 1988, [1988 Proceedings] Second International Conference on Computer Vision.

[11]  Muralidhara Subbarao,et al.  Depth recovery from blurred edges , 1988, Proceedings CVPR '88: The Computer Society Conference on Computer Vision and Pattern Recognition.

[12]  P. Grossmann,et al.  Depth from focus , 1987, Pattern Recognit. Lett..

[13]  Alex Pentland,et al.  A New Sense for Depth of Field , 1985, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[14]  Jerry M. Mendel,et al.  Lessons in digital estimation theory , 1986 .

[15]  William F. Schreiber,et al.  Fundamentals of Electronic Imaging Systems , 1986 .

[16]  Dennis C. Ghiglia.  Space-invariant deblurring given N independently blurred images of a common object , 1984 .

[17]  E. H. Linfoot.  Principles of Optics , 1961 .