Limits...
An automated calibration method for non-see-through head mounted displays.

Gilson SJ, Fitzgibbon AW, Glennerster A - J. Neurosci. Methods (2011)

Bottom Line: The centroids of the markers on the calibration object are recovered and their locations re-expressed in relation to the HMD grid.This allows established camera calibration techniques to be used to recover estimates of the HMD display's intrinsic parameters (width, height, focal length) and extrinsic parameters (optic centre and orientation of the principal ray).Our calibration method produces low reprojection errors without the need for error-prone human judgements.

View Article: PubMed Central - PubMed

Affiliation: Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK. stuart.gilson@physiol.ox.ac.uk

Show MeSH
Two components of the HMD optic centre locations, estimated for different numbers of samples for all trajectories from the 6 HMD positions. Light grey symbols represent the results when the calibration is based on only 7 samples; mid grey: 10 samples; dark grey: 20 samples; black: 100 samples. As the number of samples increases, the estimated optic centres cluster more closely around the estimate obtained using the combined trajectories for each HMD position (≈32,000 samples, origin of plot).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3142613&req=5

fig0040: Two components of the HMD optic centre locations, estimated for different numbers of samples for all trajectories from the 6 HMD positions. Light grey symbols represent the results when the calibration is based on only 7 samples; mid grey: 10 samples; dark grey: 20 samples; black: 100 samples. As the number of samples increases, the estimated optic centres cluster more closely around the estimate obtained using the combined trajectories for each HMD position (≈32,000 samples, origin of plot).

Mentions: Given that the ultimate aim is to recover the 11 projection parameters of the display, it is instructive to plot the change in these as the number of samples increases and reprojection errors drop. Fig. 8 illustrates the X and Z translation components of the optic centre plotted as a function of the number of samples used in the calibration for the 4 trajectories from the HMD position shown in Fig. 7. It can be seen that, when the sample number is low, the estimates of the optic centre are scattered around the location of the best estimate, obtained with 32,000 samples. This reduction in scatter is accompanied by a relatively modest fall in the reprojection error, from ≈2.0 pixels for 10 samples to ≈1.0 pixels for 32,000 samples. The examples illustrate the advantage of an automatic, camera-based method over those that rely on human judgements of alignment, such as SPAAM (Tuceryan et al., 2002), which are inevitably limited in the number of samples that can be obtained.


An automated calibration method for non-see-through head mounted displays.

Gilson SJ, Fitzgibbon AW, Glennerster A - J. Neurosci. Methods (2011)

Two components of the HMD optic centre locations, estimated for different numbers of samples for all trajectories from the 6 HMD positions. Light grey symbols represent the results when the calibration is based on only 7 samples; mid grey: 10 samples; dark grey: 20 samples; black: 100 samples. As the number of samples increases, the estimated optic centres cluster more closely around the estimate obtained using the combined trajectories for each HMD position (≈32,000 samples, origin of plot).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3142613&req=5

fig0040: Two components of the HMD optic centre locations, estimated for different numbers of samples for all trajectories from the 6 HMD positions. Light grey symbols represent the results when the calibration is based on only 7 samples; mid grey: 10 samples; dark grey: 20 samples; black: 100 samples. As the number of samples increases, the estimated optic centres cluster more closely around the estimate obtained using the combined trajectories for each HMD position (≈32,000 samples, origin of plot).
Mentions: Given that the ultimate aim is to recover the 11 projection parameters of the display, it is instructive to plot the change in these as the number of samples increases and reprojection errors drop. Fig. 8 illustrates the X and Z translation components of the optic centre plotted as a function of the number of samples used in the calibration for the 4 trajectories from the HMD position shown in Fig. 7. It can be seen that, when the sample number is low, the estimates of the optic centre are scattered around the location of the best estimate, obtained with 32,000 samples. This reduction in scatter is accompanied by a relatively modest fall in the reprojection error, from ≈2.0 pixels for 10 samples to ≈1.0 pixels for 32,000 samples. The examples illustrate the advantage of an automatic, camera-based method over those that rely on human judgements of alignment, such as SPAAM (Tuceryan et al., 2002), which are inevitably limited in the number of samples that can be obtained.

Bottom Line: The centroids of the markers on the calibration object are recovered and their locations re-expressed in relation to the HMD grid.This allows established camera calibration techniques to be used to recover estimates of the HMD display's intrinsic parameters (width, height, focal length) and extrinsic parameters (optic centre and orientation of the principal ray).Our calibration method produces low reprojection errors without the need for error-prone human judgements.

View Article: PubMed Central - PubMed

Affiliation: Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK. stuart.gilson@physiol.ox.ac.uk

Show MeSH