1 / 29

Motion Deblurring Using Hybrid Imaging

Motion Deblurring Using Hybrid Imaging. Moshe Ben-Ezra and Shree K. Nayar Columbia University IEEE CVPR Conference June 2003, Madison, USA. Image Recording Requires Time . Daguerre 1829 1/2 hour exposure . Niépce 1827 8 hours exposure . Motion Blur is Everywhere. Object Motion.

kacy
Download Presentation

Motion Deblurring Using Hybrid Imaging

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Motion Deblurring Using Hybrid Imaging Moshe Ben-Ezra and Shree K. Nayar Columbia University IEEE CVPR Conference June 2003, Madison, USA

  2. Image Recording Requires Time Daguerre1829 1/2 hour exposure Niépce 1827 8 hours exposure

  3. Motion Blur is Everywhere Object Motion Camera Motion

  4. Stabilized Lenses 1/15 second (< -5 stops) • Stabilization drifts with time • Rotation only 1/250 second (< -1 stop) Canon Stabilized lens 400mm

  5. Blind Image Deconvolution Accurate Point Spread Function (PSF) Needed.

  6. Motion Point Spread Function (PSF) Motion PSF is a Function of: Motion path Motion speed H Energy ~ 1/ speed Y X Spatial spread

  7. PSF Detector? PSF Detector Camera Can the PSF detector be a small and simple imaging device ?

  8. Fundamental Limits of Imaging Photon flux Detector Electron wells Pixel’s Signal Noise Detector’s noise level

  9. Fundamental Resolution Tradeoff Hybrid imaging system Hi-resolution camera Conventional video camera Low-resolution camera 3M 2048x1536 330K 720x480 75K 320x240 A Hybrid camera enjoys both worlds 3 Temporal resolution (fps) 30 130 Spatial resolution (pixels)

  10. Overview of Approach Motion Analysis y x PSF Estimation Deconvolution Low-Res. camera Same time period Hi-Res. camera

  11. Global Motion From Low Resolution Detector Objective function (Optical flow constraint) Rotation Translation Lucas Kanade

  12. Simulations: Motion Accuracy from Low- Res. Images Average Motion Error in Pixels

  13. Constraints on Continuous PSF Energy conservation constraint: • Constant flux assumption: Path is continuous and twice differentiable • Smoothness constraint:

  14. PSF Estimation from Computed Motion y y f6 f6 f1 f1 f5 f5 f2 f2 f3 f3 f4 f4 Frame 2 … Frame 5 x h4 h4 h5 h5 h3 h3 h2 h2 y h y h Frame 2 … Frame 5 Frame 2 … Frame 5 x x

  15. Deconvolution of High Resolution Image Standard iterative ratio-based algorithm* Image estimate PSF Error • Guaranties non-negative pixel result * Richardson [72] Lucy [74]

  16. Designs for Hybrid Imaging Using a beam splitter Using a special chip A rig of two cameras

  17. Our Prototype: Rig of Two Cameras Primary detector (2048x1536) Secondary detector (360x240) Resolution ratio of 1 : 36

  18. Example 1 - Blurred Hi-Res Image f = 633mm, Exp. Time 1 Sec (> -9 stops)

  19. PSF Estimation from Motion Estimated PSF 0.06 90 Y(Pixels) 10 0.001 X(Pixels) 10 130 Low resolution sequence. f = 633mm, Exp. Time 1 Sec

  20. Deblurred Image f = 633mm, Exp. Time 1 Sec

  21. Example 1 - Comparison Tripod image (Ground Truth) Deblurred image Blurred image f = 633mm, Exp. Time 1 Sec

  22. Example 2 - Blurred Night Image f = 884mm, Exp. Time 4 Sec (> -11 stops)

  23. PSF Estimation from Motion 0.003 30 Y(Pixels) 10 0.001 X(Pixels) 10 60 Low resolution sequence. f = 884mm, Exp. Time 4 Sec

  24. Deblurred Night Image f = 884mm, Exp. Time 4 Sec

  25. Example 3 - Comparison Tripod image (Ground Truth) Deblurred image Blurred image f = 884mm, Exp. Time 4 Sec

  26. Object Deblurring Problem Moving objects blend into the background

  27. Hybrid Imaging Solution (simulated) Requires clear high-resolution background image

  28. Thank you

  29. Quantifying The Affect of Motion Blur Empirical tests: RMS error. Volume of Solutions (Linear Model): Input Images High-Resolution Image Blur Decimation Uncertainty (Quantization) Volume of Solutions 1/det(A)

More Related