analysis of indica
play

Analysis of INDICA: INteraction-free DIsplay CAlibration for - PowerPoint PPT Presentation

1/15 Performance and Sensitivity Analysis of INDICA: INteraction-free DIsplay CAlibration for Optical See-Through Head-Mounted Displays Yuta Itoh and Gudrun Klinker TUM, ISMAR2014, 12 th Sep 2014 2/15 Head-Mounted Displays Non Optical


  1. 1/15 Performance and Sensitivity Analysis of INDICA: INteraction-free DIsplay CAlibration for Optical See-Through Head-Mounted Displays Yuta Itoh and Gudrun Klinker TUM, ISMAR2014, 12 th Sep 2014

  2. 2/15 Head-Mounted Displays Non Optical See-Through Optical See-Through OST-HMD

  3. 3/15 Consistency issues in OST-HMDs Temporal Visual [Zheng ’14] [Didier ’05] [Kiyokawa ’01] [Liu ’08] [Lee ’09] Social Spatial Image from Google Glass: Don't Be A [Azuma ’95] [Tuceryan ’00 ] [Genc ’02] Glasshole | Mashable

  4. 4/15 Spatial Consistency of OST-HMDs Consistent Not consistent Virtual Real

  5. 5/15 Spatial Calibration of OST-HMD Find a Projection from 3 D to 2 D 3 D 2 D Projection HMD Eye

  6. 6/15 Manual Method needs user interaction

  7. 7/15 Manual Method needs user interaction SPAAM Single Point Active Alignment Method [Tuceryan ’00 ] [Genc ’02] Projection 3 D 2 D HMD Eye

  8. 8/15 Automated Method interaction-free INDICA INteraction-free DIsplay CAlibration [Itoh ‘14]  tracks the eyeball center 3 D 2 D Projection

  9. 9/15 Problem of INDICA  Large systematic error Reprojection Error [pixel] Repeated INDICA SPAAM

  10. 10/15 INDICA needs intensive precalibration 17 DoF *† How sensitive is # of param. INDICA to the parameters? S E 0 DoF * SPAAM INDICA * Except for precalibration of the world tracking system † Full setup

  11. 11/15 Predicted Error of INDICA  E.g. screen position S is error prone E

  12. 12/15 Observed Error of INDICA  Added artificial error S to the real data E

  13. 13/15 “Eye - HMD” Camera Modeling  Is really a pinhole camera?  Human eyes are not simple • Optical and visual axes  HMD screens are not planer [Owen ‘04] • Retinal, Light-field Owen ‘04 Image by Soerfm

  14. 14/15 Summary  Automated OST-HMD calibration, INDICA, still has systematic errors  Sensitivity analysis helps us to identify the error sources  Better eye-HMD modeling is desirable

  15. 15/15 Any questions? [D04] INDICA Demo [T08] OST-HMD calibration tutorial  Jens Gurbert and Yut Itoh Slides http://stctutorial.icg.tugraz.at/ This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under PITN- GA-2012-316919-EDUSAFE.

Recommend


More recommend